Office Action Analysis — App 18729464 (public record)
Full Analysis

Office Action Response Analysis · Non-Final (CTNF)

App. No. 18/729,464

Art Unit
2142
Examiner
LANE, THOMAS BERNARD
Mailed
07/28/2026
Response period stated in the OA
“3 MONTHS from the mailing date of this communication (07/28/2026). Extensions of time may be available under 37 CFR 1.136(a), up to a maximum of SIX (6) MONTHS.”
Rejections
§112(b) ×2§102 ×1§103 ×7
Claims
20 rejected
Generated
Aug 10, 2026

The §102 missing-element arguments (ranks 1-3) are the potentially dispositive levers because a single missing limitation defeats anticipation and an examiner cannot cure a §102 rejection by combining references (MPEP § 2131) — but all three carry the same threshold caveat: the examiner's cited Abbas paragraphs (¶¶0026, 0037, 0040, 0066, 0067, 0076, 0111) were not in the retrieved excerpt and must be pulled and read before counsel treats any element as absent, so the argue-vs-amend posture on claims 1 and 18 cannot be fixed until that record gap is closed. The §103 arguments (ranks 4-5) are best framed as attacks on the articulated reasoning and reference characterization rather than strict absence, and rank 6 is the weakest given Bond's own sensor-at-intervals disclosure. Given this examiner's documented interview propensity and low average office-action count, counsel may wish to weigh whether an interview to test the §102 mappings — after pulling the cited paragraphs — is a more efficient path than a written response alone, and whether targeted amendments tying the disputed limitations (sensor-sourced measurement data, temporal fault-scenario labeling, compression-fitting decompression, the two-stage cascade) to specific specification support would provide a firmer fallback than argument alone.

Examiner Thomas Lane (AU 2142): allowance rate 82% (n=17); avg 1.29 OAs to allowance; interviews held in 29% of cases, and when an interview was held allowance followed 80% of the time (correlation, not causation); RCE filed in 29% of cases. Directional only — small sample (n=17). Based on n=24 applications; USPTO public data, 2016-01-01..2022-12-31. Correlational — it informs, it never decides.

Generated on a published USPTO office action — no confidential disclosure involved. First-pass analysis for attorney review — not a drafted response.

1.

Per-Claim Strategy

An at-a-glance recommendation per rejected claim, composed deterministically from the analysis below. A triage summary for counsel to weigh, not a decision.

ClaimRejectionsRecommended pathBasisFallback amendmentConfidence
Claim 1§102 (anticipation)ArgueMissing element (#3)high
Claim 2§112(b) (indefiniteness)§103 (obviousness)ArgueDefiniteness rebuttal (#9)moderate
Claim 3§102 (anticipation)ArgueMissing element (#3)moderate
Claim 4§103 (obviousness)ArgueMischaracterized reference (#6)high
Claim 5§103 (obviousness)Obtain reference / verify firstConclusory rationale (#12)unverified
Claims 6, 7§103 (obviousness)ArgueConclusory rationale (#8)moderate
Claim 8§102 (anticipation)ArgueMissing element (#3)moderate
Claim 9§103 (obviousness)ArgueMischaracterized reference (#5)moderate
Claim 10§102 (anticipation)ArgueMissing element (#3)moderate
Claim 11§112(b) (indefiniteness)§103 (obviousness)ArgueStrategy check: re-ranked — The current #1 for claim 11 (rank 5, Reece cascade) presupposes a claim-9-style dependency claim 11 does not currently have, and Reece supplies neural networks, so the §112b antecedent defect (rank 9) is the foundational and strongest lever for this claim.Definiteness rebuttal (#1)moderate
Claim 12§103 (obviousness)Review — no argument identifiedlow
Claim 13OtherArgueDefiniteness rebuttal (#10)moderate
Claim 14§103 (obviousness)Review — no argument identifiedlow
Claim 15§103 (obviousness)ArgueNon-analogous art (#7)low
Claim 16§103 (obviousness)Obtain reference / verify firstConclusory rationale (#12)unverified
Claim 17§103 (obviousness)ArgueNon-analogous art (#7)low
Claim 18§102 (anticipation)ArgueMissing element (#2)high
Claim 19§103 (obviousness)Obtain reference / verify firstConclusory rationale (#12)unverified
Claim 20§102 (anticipation)ArgueMissing element (#2)moderate
2.

Argument Bank

Candidate arguments for counsel, ranked strongest-first — brainstorming inputs for counsel to evaluate, not a drafted response.

1

Claim 11 antecedent basis — 'second algorithm' is introduced only in claim 9 (§112(b))

Definiteness rebuttalClaim 11Rebuts: §112(b) rejection of claim 11

Strategy check: re-ranked from #9 — The current #1 for claim 11 (rank 5, Reece cascade) presupposes a claim-9-style dependency claim 11 does not currently have, and Reece supplies neural networks, so the §112b antecedent defect (rank 9) is the foundational and strongest lever for this claim.

wherein each of the first algorithm and second algorithm comprise a neural network

For counsel to weigh: claim 11 depends from claim 8, which introduces only 'a first algorithm'; the 'second algorithm' is introduced in claim 9, which is outside claim 11's dependency chain, so the recitation 'the ... second algorithm' lacks antecedent basis. This is a well-founded formal §112(b) defect (antecedent basis, MPEP § 2173.05(e)) and is best resolved by amendment — for example redirecting claim 11's dependency so that the 'second algorithm' has proper antecedent basis. Counsel should treat this as a correction rather than a contestable substantive rejection; there is no strong argument that the scope is clear as written.

  • Claim 8: 'training a first algorithm to classify measurement data as including a pattern' (introduces only the first algorithm).
  • Claim 9: 'training a second algorithm using measurement data that is classified ... by the first algorithm' (introduces the second algorithm).
  • Claim 11: 'wherein each of the first algorithm and second algorithm comprise a neural network.'
MPEP § 2173.05(e) — lack of antecedent basis is a proper §112(b) indefiniteness ground; the customary cure is amendment.

⚠ Risk There is no persuasive counterargument that 'second algorithm' has antecedent basis in claim 8; pressing the merits would waste the response. The practical path is an amendment concept placing claim 11 in a chain that first introduces the second algorithm. No prosecution-history estoppel concern beyond the ordinary effect of the amendment.

2

Abbas trains on historical known-leak records, not 'experimental data obtained during fault scenarios' (§102, claim 18)

Missing elementClaim 18Claim 20Rebuts: §102 rejection of claims 1, 3, 8, 10, 18, 20

a leak prediction algorithm that has been trained to predict leaks based on experimental data obtained during fault scenarios that will cause a leak in future

For counsel to weigh: claim 18 was rejected under §102 over Abbas alone, yet the fully-grounded Abbas text describes training on historical known-leak pipe records, not on experimental data obtained during deliberately-introduced fault scenarios. A telling internal signal in the office action itself: the examiner drew the experimental/fault-scenario training concept from Wang (for claim 2) and from Reece (for claim 9) under §103, which suggests the examiner did not find that concept in Abbas standing alone. That is inconsistent with treating Abbas alone as anticipating the same experimental-fault-scenario training in claims 18 and 20. Caveat: the examiner's cited ¶¶0026, 0066, 0076 were not in the retrieved excerpt and should be pulled to verify.

  • Abbas Brief Summary: model trained on 'data for pipes ... and knowledge on whether those pipes leaked' (historical records).
  • OA1: for claim 2 the experimental fault-scenario data is supplied by Wang; for claim 9 by Reece — not attributed to Abbas.
  • OA3 claim chart: 'The experimental/fault-scenario training concept is the very teaching the examiner drew from Wang (claim 2) and Reece (claim 9) for other claims, suggesting Abbas alone may not supply it.'
MPEP § 2131 — §102 anticipation must rest on a single reference disclosing every element; drawing the same concept from other references under §103 undercuts a single-reference finding.

Risk The examiner may respond that Abbas's predictive framing (predicting first-time leaks) encompasses 'experimental data.' Counsel should confirm the cited paragraphs and be ready that the examiner could retreat to a §103 theory pairing Abbas with Wang/Reece. Prosecution-history caution: distinguishing on 'experimental data obtained during fault scenarios' narrows claim 18/20 scope.

Likely examiner response survives — moderate

The examiner can argue that a §102 rejection of claim 18 and §103 rejections of claims 2 and 9 (drawing the fault-scenario concept from Wang and Reece) are NOT inconsistent — different claims recite different language, and claim 18's 'experimental data obtained during fault scenarios that will cause a leak in future' may be broader or differently worded than the claim-2/claim-9 limitations, such that Abbas alone reads on claim 18 while narrower dependents needed a secondary reference. The examiner would also note ¶¶0026, 0066, 0076 are the cited basis and were not in the excerpt, and that under BRI historical records of pipes that leaked can be characterized as data 'obtained during' leak-causing conditions.

How to adjust The internal-signal point (examiner reached to Wang/Reece for the same concept elsewhere) is persuasive rhetoric but not dispositive — an examiner can legitimately reject differently-worded claims under different statutes. Confirm ¶¶0026/0066/0076 first. Strengthen by doing a side-by-side of claim 18's actual language against claims 2 and 9 to show the fault-scenario/experimental-data concept is materially the same across all three (so the §103 reliance on Wang/Reece is a tacit acknowledgment the concept is not in Abbas), and by anchoring 'experimental data obtained during fault scenarios' to a construction that excludes passively-collected historical leak records.

3

Abbas trains on pipe-characteristic records, not on environmental sensor measurement data (§102, claim 1 and dependents)

Missing elementClaim 1Claim 3Claim 8Claim 10Rebuts: §102 rejection of claims 1, 3, 8, 10, 18, 20

training measurement data from sensors monitoring an environment in proximity to the pipework

For counsel to weigh: anticipation under §102 requires that the single reference disclose every limitation arranged as claimed (MPEP § 2131). In the fully-grounded Abbas text, the training input is per-pipe CHARACTERISTIC records — the Brief Summary describes training 'using data for pipes contained in a training dataset, which may contain information regarding the characteristics of various pipes (e.g., the dimensions of those pipes, the materials of those pipes, the age of those pipes, the locations of those pipes ...),' and claim 1 recites 'the first data items include characteristics of the respective pipes.' The available Abbas text does not describe environmental sensors, or sensor measurement data monitoring an environment in proximity to the pipework, as the training input. Because claims 3, 8, and 10 depend from claim 1, this candidate absence, if it holds, reaches all four claims. One caveat counsel must resolve first: the examiner's cited support (¶¶0067, 0111) was not present in the retrieved excerpt, so those specific paragraphs should be pulled and read before this element is treated as absent.

  • Abbas Brief Summary: 'The predictive model may be trained using data for pipes contained in a training dataset, which may contain information regarding the characteristics of various pipes (e.g., the dimensions of those pipes, the materials of those pipes, the age of those pipes, the locations of those pipes, and so forth) and knowledge on whether those pipes leaked.'
  • Abbas claim 1: 'the first data items include characteristics of the respective pipes.'
  • OA3 claim chart: Abbas record training input is 'pipe CHARACTERISTIC records ... the available fully-grounded text does not describe environmental sensors or sensor measurement data as the training input.'
MPEP § 2131 — §102 requires a single reference to disclose every element arranged as in the claim; an examiner cannot fill a gap by combining references.

⚠ Risk The examiner will likely point to ¶¶0067/0111 (not in the retrieved excerpt) as describing sensors monitoring the pipe and environment; counsel should pull those paragraphs before pressing this, because if they do describe environmental sensors the argument weakens. Prosecution-history caution: characterizing 'environmental sensors' as the distinguishing feature narrows claim scope to sensor-measurement training data and may create estoppel against a later broader reading.

Likely examiner response survives — moderate

The examiner can point out that the rejection expressly relies on Abbas ¶¶0067 and 0111 — paragraphs that were NOT in the retrieved excerpt — for the sensor-measurement-input mapping, so the absence argument rests on an incomplete reading of the reference. The examiner would maintain that Abbas discloses a broad supervised-ML pipeline that, in the un-retrieved paragraphs, may describe sensor-sourced inputs, and that Abbas's listed learning techniques (random forest, neural networks, etc.) are agnostic to input source. Under BRI the examiner can also argue that 'training measurement data from sensors monitoring an environment in proximity to the pipework' is not so narrow that pipe-condition/attribute data logged over an area could not read on it, absent a claim definition confining 'environment' and 'measurement data.'

How to adjust This argument cannot be pressed until ¶¶0067 and 0111 (and 0167-0168) are pulled and read — if they disclose sensor measurement data as the training input, the missing-element premise collapses. Shore it up by (a) confirming from the full Abbas text that the training input is limited to pipe-characteristic/attribute records from a pipe database, and (b) if 'environment' and 'sensor measurement data' are being read broadly, tightening the specification-based construction that distinguishes environmental sensor streams from static pipe-attribute records. If the pulled paragraphs supply sensor data, pivot to amendment emphasizing the temporal/environmental sensing distinction.

4

Abbas discloses per-pipe known-leak labels, not fault-scenario period labels (§102, claim 1)

Missing elementClaim 1Claim 3Claim 8Claim 10Rebuts: §102 rejection of claims 1, 3, 8, 10, 18, 20

labels indicating which periods of the training measurement data correspond with one or more fault scenarios selected to cause leaks in future

For counsel to weigh: the fully-grounded Abbas text labels each pipe with 'known leaks' — a binary record that a given pipe leaked or did not — rather than labels marking which time-periods of measurement data correspond to deliberately-introduced fault scenarios selected to cause leaks in future. Abbas claim 1 recites 'known leaks associated with respective pipes,' and the Detailed Description frames the labels as 'knowledge on whether those pipes leaked.' The claimed limitation requires temporal labeling tied to introduced fault scenarios, a concept the Abbas record does not appear to reach. This candidate absence, combined with the missing sensor-measurement input (rank 1), goes to the heart of the §102 mapping. Caveat: the examiner's cited ¶¶0037, 0040 were not in the retrieved excerpt and must be pulled to confirm.

  • Abbas claim 1: 'a training dataset including first data items and known leaks associated with respective pipes of a first plurality of pipes.'
  • Abbas Detailed Description: training data includes 'knowledge on whether those pipes leaked.'
  • OA3 claim chart: 'The fully-grounded Abbas text discloses per-pipe known leaks labels (a given pipe leaked or did not), not labels marking which time-periods of measurement data correspond to deliberately-introduced fault scenarios.'
MPEP § 2131 — every claim element, including the label limitation, must be present in the single reference; MPEP § 2112 — inherency requires necessity, not mere possibility.

Risk The examiner may argue that Abbas's 'causes of leaks' data (cited ¶¶0037, 0040) inherently teaches fault-scenario labeling; counsel should confirm those paragraphs do not describe deliberate fault introduction with periodic labels. Prosecution-history caution: emphasizing the 'periods ... fault scenarios selected to cause leaks' construction narrows scope and may estop a broader label reading later.

Likely examiner response survives — moderate

The examiner can respond that Abbas ¶¶0037 and 0040 — again not in the retrieved excerpt — are the cited basis for the labeling limitation, and that under BRI a 'known leaks' label that records when a pipe failed can be characterized as indicating periods of data corresponding to leak conditions. The examiner may argue the claim does not require that the fault scenarios be 'deliberately introduced' in any structurally distinct way beyond a label tied to leak-causing conditions, and that the phrase 'selected to cause leaks in future' is aspirational/intended-use language that does not further limit the training-data structure Abbas discloses.

How to adjust Confirm ¶¶0037/0040 before relying on the absence. The distinguishing feature is temporal labeling of measurement-data periods tied to introduced fault scenarios — versus a per-pipe binary leaked/not-leaked record. Strengthen by anchoring 'labels indicating which periods' to a specification passage that requires time-indexed labeling of a measurement-data stream, so the examiner's binary-record reading is foreclosed under a proper construction. This argument is analytically paired with rank 1 (same sensor-measurement-input premise); if rank 1's premise survives the paragraph pull, this one likely does too, and vice versa.

5

Reece teaches leak/no-leak detection, not a two-algorithm classify-then-train cascade or leak-type classification (§103, claims 9 and 11)

Mischaracterized referenceClaim 9Claim 11Rebuts: §103 rejection of claims 9, 11

training a second algorithm using measurement data that is classified as having a pattern by the first algorithm

For counsel to weigh: Reece is fully grounded, and its core approach is training a computer system on a first (no-leak) dataset and a second (simulated-leak) dataset and communicating which periods had leaks — a leak/no-leak determination. Reece claim 1 recites 'training the computer system to detect the leaks ... communicating ... that no leaks existed ... and ... that leaks existed.' The examiner's stated motivation — to 'predict what kind of leak will occur' via 'multiple classification models' — does not track Reece, whose stated framing is a leak/no-leak alarm rather than classifying leak types. Nor does the retrieved Reece text describe the claimed cascade in which a first algorithm classifies data as having a pattern and a second algorithm is then trained on that classified data. Counsel should press whether the examiner read a two-stage architecture and leak-type classification into Reece that its text does not support.

  • Reece claim 1: 'training the computer system to detect the leaks in the pipeline including communicating to the computer system that no leaks existed while the first set of data was acquired and communicating to the computer system that leaks existed while the second set of data was acquired.'
  • OA2: 'The invention's stated framing is detecting/alarming on leaks (a leak / no-leak determination), not classifying leak types.'
  • OA1: examiner motivation for claims 9/11 is 'to not only predict if a leak will occur but also predict what kind of leak will occur.'
MPEP § 2145 / § 2143.01 — a reference must be read as it actually teaches; a motivation resting on a teaching the reference does not contain is conclusory.

Risk The examiner may point to Reece ¶0064 (cited but within the truncated Detailed Description) as describing multiple classification models; counsel should pull ¶0064 and ¶0013 to confirm they do not disclose the claimed classify-then-train cascade before relying on this. Note claim 11's separate antecedent-basis defect (see the §112 candidate).

Likely examiner response survives — moderate

As a §103 rejection, the examiner can argue Reece need not disclose the two-stage classify-then-train cascade or leak-type classification outright — Reece's training on a first (no-leak) dataset and a second (simulated-leak) dataset provides the building blocks, and combining/extending them to a multi-model or staged architecture to 'predict what kind of leak will occur' is a predictable use of a known technique (KSR (C)/(D)). The examiner would resist an attack on Reece in isolation where the rejection rests on a combination, and can characterize Reece's two-dataset framework as suggesting sequential/multi-model processing.

How to adjust Keep the attack on the combination and the articulated motivation, not on Reece alone (MPEP § 2145). The pressable gap: Reece's stated framing is a leak/no-leak alarm, and the claimed cascade requires a first algorithm to classify data as having a pattern and a second algorithm trained on that classified output — a specific architecture the retrieved Reece text does not describe. Press whether the examiner's 'multiple classification models' / leak-type motivation has any record support in Reece or is conclusory. Strengthen by mapping the claimed cascade element-by-element to show two static datasets are not the same as a trained-classifier feeding a second trainer; if Reece's Detailed Description (truncated in the record) is pulled and still lacks the cascade, that reinforces the gap.

6

Mezghani discloses a solder/PVC-cement sealed joint, not 'progressive decompression of a compression fitting' (§103, claim 4)

Mischaracterized referenceClaim 4Rebuts: §103 rejection of claims 4, 12, 14

the one or more fault scenarios comprises progressive decompression of a compression fitting

For counsel to weigh: Mezghani is fully grounded, and its disclosed embodiments form the pipe joint 'with solder or PVC cement,' explaining that 'Over a period of time, the seal may deteriorate so that the pipes leak.' That is seal deterioration of a soldered/cemented joint — not a compression fitting, and not a 'progressive decompression.' Whether solder/PVC-cement seal deterioration reads on 'progressive decompression of a compression fitting' is a genuine gap for counsel to press, because the examiner (¶0016) appears to have equated a deteriorating solder/cement seal with a decompressing compression fitting. The two are structurally and mechanically distinct fault mechanisms, which bears on both the mapping and the motivation to combine.

  • Mezghani Detailed Description: 'an exemplary pipe joint is formed between a pipe P2 having a fitting F and a second pipe P1 seated in the fitting F, with solder or PVC cement forming a seal in the fitting.'
  • Mezghani: 'Over a period of time, the seal may deteriorate so that the pipes leak.'
  • OA3 claim chart: 'a joint sealed with SOLDER or PVC CEMENT whose seal may deteriorate over time — not a compression fitting, and not a progressive decompression.'
MPEP § 2143.01 / § 2145 — the rejection must map the reference's actual teaching; a mischaracterized teaching cannot supply the missing limitation or a rational motivation to combine.

Risk The examiner may argue that seal deterioration and progressive decompression are functionally similar loss-of-seal mechanisms and that 'compression fitting' reads broadly. Counsel should be prepared to define 'compression fitting' and 'progressive decompression' by reference to the specification. Prosecution-history caution: a narrowing definition of these terms will bind in later interpretation.

Likely examiner response survives — moderate

Because this is a §103 rejection (not §102), the examiner does not need Mezghani to disclose 'progressive decompression of a compression fitting' verbatim — only a rational reason a PHOSITA would arrive at it. The examiner can argue Mezghani's joint, where 'a first pipe is seated in a fitting of a second pipe,' is itself a fitting-type coupling, and that seal deterioration causing a gradual-onset leak is the same general class of progressive joint failure; substituting one known joint-failure fault mechanism for another to yield a predictable leak is a simple design choice / KSR (A)-(B) rationale. The examiner (¶0016) can maintain the mechanisms are close enough that the modification is predictable.

How to adjust Frame this as an obviousness attack, not a strict missing-element attack — the strongest form challenges the articulated reasoning: a solder/PVC-cement seal deteriorating is structurally and mechanically a different fault than a compression fitting losing compression, so counsel can press that the examiner substituted a different fault mechanism without a rational underpinning (MPEP § 2143.01) and possibly with hindsight. Confirm the ¶0016 mapping. If the specification defines 'compression fitting' and 'progressive decompression' as a specific mechanism absent from Mezghani, that construction is the lever; if the examiner has a plausible simple-substitution rationale, amending claim 4 to tie the fault scenario to compression-fitting-specific structure may be the cleaner path.

7

Bond is directed to equipment-deployment mechanics, raising a non-analogous-art question (§103, claims 15 and 17)

Non-analogous artClaim 15Claim 17

the sensors are distributed in different locations about the pipework / spaced apart by a distance of at least 50 centimeters (cm)

For counsel to weigh: Bond is fully grounded and is directed to an apparatus and method for deploying equipment into pressurized fluid mains using a piston-driven fluid housing and an anti-buckling winch — 'deployment mechanics, not ... leak detection algorithms or ... specifying inter-sensor spacing distances' (OA2). Under MPEP § 2141.01(a), counsel can test whether Bond is in the same field of endeavor as a computer-implemented leak-prediction training method and whether it is reasonably pertinent to the inventor's problem of arranging environmental sensors for leak prediction. Bond does mention that 'the equipment comprises a sensor, such as an acoustic sensor' and that 'a number of items of equipment each mounted at intervals along the carrier' may be used, so the examiner may argue pertinence; but the retrieved Bond text is about how to push equipment into a main, not where to space leak-detection sensors. Counsel should also confirm the examiner's cited ¶0091 (a specific 1-meter spacing) is actually in Bond and describes sensor spacing rather than deployment intervals.

  • OA2: Bond 'is directed to deployment mechanics, not to leak detection algorithms or to specifying inter-sensor spacing distances.'
  • Bond Description: 'the equipment comprises a sensor, such as an acoustic sensor ... It is possible to use a number of items of equipment each mounted at intervals along the length of the carrier.'
  • Bond abstract/claims: apparatus for 'deploying equipment into a fluid container' via a piston and guide means.
MPEP § 2141.01(a) — a reference supports §103 only if same field of endeavor or reasonably pertinent to the inventor's problem.

Risk The examiner will likely argue Bond concerns water-main sensors and is at least reasonably pertinent to sensor placement, and that spacing 'items ... at intervals' meets the limitation. Counsel should verify ¶0091 exists and describes inter-sensor spacing; if it does, the non-analogous-art angle weakens and the argument becomes whether deployment intervals equate to the claimed sensor spacing.

Likely examiner response fragile — the comeback likely defeats it

The examiner has a real answer on both analogous-art prongs: Bond concerns deploying equipment into pressurized water mains and expressly states the deployed 'equipment comprises a sensor, such as an acoustic sensor' mounted 'at intervals along the carrier,' with cited ¶0091 reciting a specific spacing. The examiner can argue Bond is in the same field of endeavor (sensing within pipework/water mains) and, in any event, is reasonably pertinent to the inventor's problem of positioning/spacing leak-detection sensors about the pipework — the very problem claims 15 and 17 address (MPEP § 2141.01(a)). Non-analogous-art challenges are difficult where the reference explicitly discusses sensors placed at intervals in a main.

How to adjust Non-analogous-art is a hard win here because Bond's own text supplies sensors deployed at spaced intervals in a water main, giving the examiner a plausible field-of-endeavor AND reasonably-pertinent response. Do not lead with this. First confirm whether ¶0091 actually recites inter-sensor spacing for leak detection or merely deployment intervals for pushing equipment along a carrier — that factual distinction is the only strong lever. If ¶0091 describes deployment spacing rather than leak-detection sensor spacing, reframe as a mischaracterized-reference / no-teaching attack on the specific 50 cm limitation and its motivation, rather than a pure analogous-art argument; otherwise consider amending claims 15/17 toward the sensor-arrangement feature that Bond does not reach.

8

Davis mapping for the 30 cm / enclosed-cavity limitations rests on a paragraph not in the record (§103, claims 6 and 7)

Conclusory rationaleClaim 6Claim 7Rebuts: §103 rejection of claims 6, 7

the sensors are placed within 30 centimeters (cm) of the pipework / at least some of the sensors are placed in an enclosed cavity with the pipework

For counsel to weigh: Davis is fully grounded at the claim/abstract level, but its Detailed Description 'is cut off well before the paragraph the examiner relies on' (¶0321), so the specific factual support for 'within 30 cm' and for an 'enclosed cavity' is not verifiable in the retrieved text. The retrieved Davis claims describe a containment apparatus 'at least partially encapsulating' the conduit with a sensor unit 'associated with the containment apparatus' — which may support an enclosed-cavity reading for claim 7, but the record does not show an express 30 cm distance for claim 6. Counsel should require the examiner to identify where Davis discloses the specific 30 cm proximity rather than assuming it from a figure, and weigh whether the motivation ('more accurately detect leaks with less interference') is articulated beyond a conclusory statement.

  • Davis claim 1: 'a containment apparatus sealingly attachable to, and at least partially encapsulating, the fluid conduit system ... a sensor unit associated with the containment apparatus.'
  • OA2: 'the description is cut off well before the paragraph the examiner relies on.'
  • OA1: examiner relies on Davis ¶0321 and 'fig. 1A-B show the sensor touching the pipe' for the 30 cm limitation.
MPEP § 2143.01 — the rejection must articulate factual support; a distance limitation cannot rest on an unverified paragraph or an inference from a figure.

Risk The examiner will likely reproduce ¶0321 and Fig. 1A-B showing a sensor touching the pipe, which would readily satisfy 'within 30 cm' and 'enclosed cavity.' This is a verify-the-support point more than a substantive distinction; counsel should confirm ¶0321 before over-investing.

9

'Period of time' in claim 2 — scope may be reasonably certain under the examination standard (§112(b))

Definiteness rebuttalClaim 2Rebuts: §112(b) rejection of claim 2

so as to cause leaks at a period of time after the introduction of the one or more fault scenarios

For counsel to weigh: the examiner rejected 'period of time' as a relative term of indefinite scope. The correct examination lever is MPEP § 2173.02 / In re Packard / Ex parte Miyazaki — whether the term is unclear or amenable to more than one plausible construction under the broadest reasonable interpretation — not the litigation 'reasonable certainty' standard of Nautilus. Counsel can argue that, read in context, 'a period of time after the introduction of the one or more fault scenarios' simply denotes that leaks occur some elapsed time after the fault is introduced (i.e., the fault precedes the leak), which is a determinate temporal relationship rather than an indefinite degree term. Counsel should check the as-filed specification for any language giving temporal context; if the scope cannot be shown clear under BRI, an amendment clarifying the temporal relationship is the alternative path.

  • OA1 §112(b) rejection: 'The term period of time renders the cause of a leak indefinite because it is not clear as to when the leak can appear.'
  • Claim 2: 'so as to cause leaks at a period of time after the introduction of the one or more fault scenarios.'
MPEP § 2173.02 — during examination definiteness is tested under In re Packard (unclear term) / Ex parte Miyazaki (multiple plausible constructions), not the Nautilus litigation standard.

Risk The examiner may maintain that the term states no lower/upper temporal bound and remains indefinite. Cite § 2173.02 / In re Packard — do NOT cite Nautilus, which would invite correction. If the specification lacks supporting context, amendment is the more reliable route.

10

Claim 13 improper multiple dependent form — conjunctive rather than alternative (§112(e))

Definiteness rebuttalClaim 13

The method of claim 12 ... trained according to the method of claim 1

For counsel to weigh: the examiner rejected claim 13 as an improper multiple dependent claim referring to two claims in the conjunctive rather than the alternative (37 CFR 1.75(c) / § 112(e)). As-listed, claim 13 depends only on claim 12 and recites training 'according to ... claim 1,' so counsel should first confirm against the actual file wrapper whether claim 13 in fact refers to two claims conjunctively, since the pending claim listing shows a single dependency on claim 12. If the examiner's reading is correct, this is a formal defect curable by amendment to a proper single or alternative dependency; there is no substantive counterargument to the form requirement. Counsel should also note the examiner stated claim 13 'cannot be further examined on its merits,' so curing the form is necessary to obtain a merits examination.

  • OA1: claim 13 rejected under §112(e) because it 'refers to two claims in the conjunctive (claim 12 AND claim 1) rather than the alternative.'
  • Pending claim 13 listing: 'The method of claim 12, wherein the leak prediction algorithm has been trained according to the method of ... claim 1.'
MPEP § 608.01(n) / 37 CFR 1.75(c) — a multiple dependent claim must refer to preceding claims in the alternative; the form defect is cured by amendment.

Risk The form requirement is not contestable on the merits; the response should present an amendment concept. Confirm the actual filed claim text, since the pending listing appears to show only a single dependency on claim 12 — a discrepancy worth clarifying with the examiner.

11

Wang (§103, claim 2) is unverifiable — obtain and read before relying on any distinction

Conclusory rationaleClaim 2Rebuts: §103 rejection of claim 2

the training measurement data is obtained during controlled experiments in which the one or more fault scenarios are introduced

For counsel to weigh (verify-first only): the Wang NPL reference was never retrieved into the record, so its text cannot be checked against the examiner's characterization that its section 2/section 5 experimental platform introduces faults and trains a prediction model on the resulting data. Because the reference text is unavailable, counsel cannot presently establish that any limitation is missing from Wang or that the examiner mischaracterized it; the only responsible posture is to obtain Wang and verify (a) that its experimental fault-introduction and training disclosure is as the examiner describes and (b) whether the articulated motivation ('to train the algorithm on data for leaks that might not have a lot of data') is supported. This candidate must rank below every argument resting on a fully-grounded reference.

  • Reference grounding: Wang not listed among grounded references (text not retrieved).
  • OA1: examiner relies on Wang 'page 3, section 2' and 'section 5' for controlled-experiment fault introduction and training.
MPEP § 2143 — the combination rationale requires articulated factual support; that support cannot be evaluated until the reference is in the record.Evidence needed: Obtain and read the Wang reference (page 3, sections 2 and 5) to confirm what its experimental platform and training disclosure actually teach before framing any distinction.

Risk Do not assert Wang lacks any teaching until the text is in hand — a negative assertion about an unread reference is unsupportable and risks being contradicted once Wang is produced. Ask the examiner to make Wang of record.

12

Kamkalow (§103, claims 5, 16, 19) is unverifiable — the number did not resolve to a document

Conclusory rationaleClaim 5Claim 16Claim 19Rebuts: §103 rejection of claims 5, 16, 19

the sensors monitoring the environment comprise at least one of: a humidity sensor, a temperature sensor, and an atmospheric pressure sensor

For counsel to weigh (verify-first only): the reference cited as 'Kamkalow (WO9941580A1)' did not resolve to a retrievable document, so its content is known only through the examiner's characterization that it discloses temperature, humidity, and pressure sensors for pipe leak detection. Counsel cannot presently confirm that characterization, the reference's publication status/date, or its analogous-art posture. The responsible step is to obtain the actual document (and confirm the correct publication number) before either accepting the mapping to the environmental-sensor limitations of claims 5, 16, and 19 or arguing any distinction. This candidate ranks below all fully-grounded arguments.

  • Reference grounding: 'Kamkalow (WO9941580A1): UNVERIFIABLE — this number did not resolve to a document — verify it.'
  • OA1: examiner relies on 'Kamkalow, page 3, paragraph 4' for temperature, humidity and pressure sensors.
MPEP § 2143 / § 2141.01(a) — the reference must be identified, of record, and analogous; a rejection cannot rest on an unretrievable document without counsel's verification.Evidence needed: Locate the correct Kamkalow publication and obtain its text (page 3, paragraph 4) to confirm the disclosed sensors and its date/status before relying on any distinction.

Risk Do not argue the reference lacks a teaching before reading it. First confirm the correct publication number and obtain the document; if it cannot be located, counsel may request that the examiner provide a copy and a correct citation.

3.

Examiner's Characterization of the Cited Art

Note

What each cited reference actually discloses, checked against what the examiner said it teaches — limited to the reference text available to the analysis.

Abbas (US 2018/0300639 A1)

US 20180300639 A1Claim text retrieved

The available Abbas text describes a computer-implemented pipe leak prediction system that trains a supervised machine learning predictive model (e.g., random forest, logistic regression, naive Bayes, and other listed techniques including neural networks) on a training dataset of pipe 'characteristics' (dimensions, materials, age, location, contents) together with 'known leaks' labels, validates the model with a confusion matrix and true/false positive/negative rates, and then applies the model to predict which pipes will leak, including predicting when a pipe will fail. In the available text, the training data source is pipe characteristic/attribute records supplied by a pipe installer and stored in a pipe database — the available text does not describe environmental sensors or sensor measurement data as the training input. The full numbered specification (including the specific paragraphs the examiner cites, e.g. ¶¶0067, 0111, 0167-0168) is not present in the available excerpt and should be pulled to verify the paragraph-specific assertions.

Claim elementExaminer assertsReference disclosesEvidence
Claims 1, 18, 20, 12 — input is training/measurement data from sensors monitoring an environment in proximity to the pipework (claim 18: 'a plurality of environmental sensors disposed in proximity to the pipework')Abbas ¶¶0067, 0111 teach data that 'comes from sensors that are monitoring the pipe and the environment.'Not found in available textThe available text (Abstract, claims 1-20, Detailed Description excerpt) does not contain any reference to environmental sensors or to sensor measurement data as the training input. Instead the available text affirmatively describes the training input as pipe attribute records — the installer provides "information about the location of the pipe, the dimensions and material of the pipe, the purpose and contents carried by the pipe" (Abbas), and claim 1 defines the data items as "characteristics of the respective pipes." The specific cited paragraphs ¶¶0067 and 0111 are beyond the available excerpt and should be pulled to confirm whether they actually describe environmental sensors; on the face of the available text the data source is materially different from sensor-based environmental monitoring. For counsel to weigh (§102 / MPEP §2131 — anticipation requires each element).
Claims 1, 20 — labels indicating which periods of the training measurement data correspond with one or more fault scenarios selected to cause leaks in futureAbbas ¶¶0037, 0040 teach data containing information about pipes that have leaked and the causes of the leaks used to train the algorithm.Partially supportedThe available text supports labeled training on leaks: claim 1 recites "known leaks associated with respective pipes," and the Description discusses determining "the underlying cause of pipe leaks" and "variable importance." But the available text describes labels as per-pipe 'known leak' status, not labels marking 'which periods of the training measurement data' correspond to leaks, and does not describe 'fault scenarios selected to cause leaks in future' (deliberately introduced faults). Notably, the examiner turned to Wang for the 'controlled experiments / introduced fault scenarios' aspect (claim 2), which is consistent with this fault-scenario/period-labeling concept not being found in Abbas's available text. For counsel to weigh the 'periods' and 'fault scenarios selected to cause leaks' language against Abbas (§102 element-by-element).
Claims 12, 14 — providing/transmitting the measurement data to a computer via a networkAbbas ¶¶0167-0168 teach the system transmitting the measurement data to the computer system.Not found in available textThe available text does not contain ¶¶0167-0168 (the description is truncated before that point) and does not, in the available portion, expressly describe transmitting measurement data over a network; it describes an installer providing data via a computing device to a pipe integration system and generally references distributed/cloud configurations. The cited paragraphs should be pulled to verify. This also carries the same 'measurement data' framing concern noted for the sensor-input element.
Claims 1, 20 — supervised training comprises adjusting parameters to improve the accuracy of the predictionAbbas ¶¶0036, 0037, 0040 teach updating the predictive model over time as new data is added and updating model parameters to improve prediction.Partially supportedThe available text supports training a model and measuring/improving accuracy: validation "to determine an accuracy of the leak predictions," generation of "a confusion matrix," and thresholds on "true positive rate" / "true negative rate" (Abbas claim 1); the Description notes the pipe database "will change and be updated over time." However, the specific '¶0036/0040' statements about iteratively 'adjusting parameters' as new data is added are not verbatim in the available excerpt; the express notion of parameter adjustment is inferential from supervised training in the available text. For counsel to weigh whether the cited paragraphs supply the specific language.
Claim 8 — training a first algorithm to classify measurement data as including a patternAbbas ¶0037 teaches an algorithm that classifies patterns related to leak detection using measurement data.Partially supportedDescription: classification techniques "classify each pipe into one or more categories (e.g., likely to leak or not likely to leak)" and the model determines "patterns associated with pipes with leaks" (Abbas). Abbas discloses classification and pattern determination, but frames it as classifying pipes into leak/no-leak categories rather than classifying 'measurement data as including a pattern'; the specific 'measurement data' framing overlaps with the sensor-input issue above. For counsel to weigh.
Claim 1 / 20 — performing supervised training of the leak prediction algorithmAbbas ¶¶0067, 0111 teach training a (random forest) leak prediction algorithm.SupportedAbbas claim 1: "applying a supervised machine learning technique to generate a predictive model ... by training the predictive model"; Abstract: "a predictive model can be trained using data"; dependent claim 2 recites a "random forest model." The general supervised-training concept is clearly present in the available text.
Claims 1, 18, 20 — output is a prediction of whether a leak is likely to occur in futureAbbas ¶¶0026, 0066, 0076 teach predicting whether a pipe will leak in the future without prior knowledge of a past leak.SupportedAbstract: predictive model "can be applied to data for various pipes in order to predict which of those pipes will leak"; Description: system is "useful for predicting the first occurrence of a leak" and can "predict not only when a pipe will fail, but also at what time it will fail" (Abbas). Future-leak prediction is well supported in the available text.
Claim 3 — the method produces an algorithm that predicts leaks prior to them becoming significantAbbas ¶0032 teaches predicting leaks before they happen or become significant.SupportedDescription: utility companies benefit from "being able to predict in advance which pipes will leak, as that would allow the companies to devote their resources to stopping these leaks before they happen"; system can "predict ... at what time it will fail" (Abbas). Prediction in advance is supported; the specific phrase 'becoming significant' is not verbatim but the advance-prediction concept is present.
Claim 10 — the leak prediction algorithm comprises an artificial neural networkAbbas ¶¶0030-0031 teach the algorithm can be a neural network.SupportedDescription lists supervised techniques "including ... artificial neural network" and classification-type techniques including "neural networks" (Abbas).

Mezghani

US 20110227721 A1Claim text retrieved

Mezghani discloses a hardware leak-detection system for a pipe joint or coupling: a transmitter is mounted adjacent a joint formed where a first pipe is seated in a fitting of a second pipe, and a pair of spaced contacts sit in an annular recess. When electrically conductive liquid (e.g., water) leaks from the joint, it bridges the gap between the contacts, closing a circuit that activates the transmitter to send an alert signal to a remote receiver. The sealed joint in the disclosed embodiments is formed with solder or PVC cement, and the reference explains that 'Over a period of time, the seal may deteriorate so that the pipes leak.' The system is reactive (it detects an actual leak) rather than predictive.

Claim elementExaminer assertsReference disclosesEvidence
Claim 4 — the one or more fault scenarios comprises progressive decompression of a compression fittingMezghani (¶0016) teaches detecting pipe leaks at the point of a fitting that is compressed between two pipes, caused by seal deterioration (i.e., progressive decompression).Partially supportedThe available text supports leaks at a fitting caused by gradual seal deterioration: "Over a period of time, the seal may deteriorate so that the pipes leak." However, the disclosed seal is formed "with solder or PVC cement," and the reference nowhere uses the terms 'compression fitting,' 'compression,' or 'decompression.' The examiner's parenthetical equating 'seal deterioration' with 'progressive decompression of a compression fitting' is the examiner's characterization, not language found in the reference — for counsel to weigh whether a soldered/cemented joint reads as a 'compression fitting' and whether seal deterioration reads as 'progressive decompression.'
Claim 12 — providing an alert in the event a leak is predictedMezghani (¶0016) teaches the sending of an alert when a leak is detected.SupportedAmply supported for the narrow proposition of sending an alert on leak detection: the Abstract states the system generates "an alert signal when a liquid leak from the pipe joint or coupling is detected," and claim 1 recites "activating the transmitter to transmit an alert signal to a remote receiver warning of a leak." Counsel may note, however, that Mezghani alerts upon an ACTUAL detected leak (conductive liquid bridging the contacts), not upon a prediction — a distinction relevant to claim 12's 'responsive to an output from the leak prediction algorithm... in the event a leak is predicted.'

Davis

US 20160284193 A1Claim text retrieved

Davis discloses a fluid leak detection apparatus in which a containment apparatus is sealingly attached to and at least partially encapsulates a fluid conduit system, capturing leaking fluid before it contacts surrounding structure, with a sensor unit associated with the containment apparatus and an alarm that generates a notification. The available text describes sensors positioned adjacent a fluid accumulation area within a container/sleeve that encloses at least a portion of the pipe, in both 'open' and 'enclosed' configurations for retrofit or new-build installation. The available text is limited to the abstract, the claims, and a truncated portion of the description (background/summary); the description is cut off well before the paragraph the examiner relies on.

Claim elementExaminer assertsReference disclosesEvidence
Claim 6 — sensors placed within 30 centimeters (cm) of the pipeworkDavis ¶0321 teaches a sensor affixed to a container that is attached to the pipe, wherein Fig. 1A-B show the sensor touching the pipe.Not found in available textThe available text does not contain ¶0321 or Figs. 1A-B — the description is truncated in the background/summary long before ¶0321, and no figures are provided. The available text nowhere states a numeric distance (e.g., 30 cm) or that the sensor touches the pipe; it recites only that a 'sensor unit [is] associated with the containment apparatus' (claim 1) and 'positioned adjacent the fluid accumulation area' (claim 7). The full specification and figures should be checked to verify the ¶0321/Fig. 1A-B assertion.
Claim 7 — at least some of the sensors placed in an enclosed cavity with the pipeworkDavis ¶0321 teaches a sensor affixed to an enclosed container (i.e. cavity) that is attached to the pipe.Not found in available textThe specific cited support (¶0321) is not present in the available text (truncated description; no figures). However, related disclosure in the available text is consistent with the concept: the summary describes an 'enclosed configuration,' claim 5 recites 'a sleeve to enclose at least a portion of the fluid conduit system,' and claim 20 recites 'a container ... at least partially enclosing at least a portion of a fluid conduit system' with 'a sensor positioned adjacent a sloped aspect of the container.' Whether ¶0321 specifically supports the 'enclosed cavity with the pipework' mapping should be verified against the full specification and figures.

Reece

US 20210216852 A1Claim text retrieved

The available text (abstract, claims 1-20, and portions of the specification including Related Applications, Field, Background, Summary, and Brief Description of Drawings, with the Detailed Description truncated) teaches a computer-implemented AI/deep-learning method of detecting leaks in a pipeline conveying liquid or gas. The core approach is training a computer system on a first data set gathered during normal (no-leak) operation and a second data set gathered while simulating leaks, then communicating to the system which periods had leaks. The disclosed models are predominantly neural-network based, including deep learning models, recurrent neural networks (RNN), and Long Short Term Memory (LSTM) RNNs. The invention's stated framing is detecting/alarming on leaks (a leak / no-leak determination), not classifying leak types.

Claim elementExaminer assertsReference disclosesEvidence
Claim 9: training a second algorithm using measurement data that is classified as having a pattern by the first algorithmReece ¶0064 teaches the use of a multiple classification models to be able to predict leaks and classify leaks based on measurement data.Not found in available textThe available text does not contain ¶0064 (the Detailed Description is truncated) and provides no paragraph numbering. Within the portions available (abstract, claims, Summary, Background), there is no affirmative disclosure of a two-stage arrangement in which a first classifier labels/classifies data that is then used to train a distinct second algorithm; the disclosed training uses a single system trained on no-leak and simulated-leak data. The only 'multiple network' language (claim 15) describes copies of a same network passing messages (RNN unrolling), not a first-then-second classifier pipeline. The full specification, including ¶0064, should be pulled and checked.
Claim 11: each of the first algorithm and second algorithm comprise a neural networkReece ¶¶0013, 0064 teaches the use of a multiple classification models, which can be neural networks, to be able to predict and classify leaks based on measurement data.Partially supportedThe neural-network predicate is strongly supported in the available text: claim 6 ("using neural networks"), claim 12 ("recurrent neural networks"), claim 16 (LSTM RNNs), and the Summary ("train an AI or Deep-Learning platform"; "using deep learning models, using neural networks"). However, the compound predicate the element requires — a FIRST and a SECOND algorithm each being a neural network — depends on Reece disclosing two distinct classification models, and that 'multiple classification models' teaching is not found in the available text (cited ¶¶0013, 0064 are in the truncated Detailed Description). The available text shows neural networks used within what is described as a single leak-detection system.

Bond (US 20020148294 A1)

US 20020148294 A1Claim text retrieved

The available text describes an apparatus and method for deploying equipment into a fluid container or conduit (especially pressurized water mains) using a fluid housing with an outlet aperture, a piston dividing the housing into two fluid chambers, and a guide means/carrier driven by a fluid-pressure differential to push equipment into the container without releasing pressure. It also describes a winch assembly for feeding an elongate carrier without buckling. The 'equipment' deployed can be a sensor, such as an acoustic sensor, or a transmitter, and multiple items of equipment can be mounted at intervals along the carrier. The reference is directed to deployment mechanics, not to leak detection algorithms or to specifying inter-sensor spacing distances.

Claim elementExaminer assertsReference disclosesEvidence
Claim 17 — sensor units spaced apart by a distance of at least 50 centimetersBond ¶0091 teaches sensors for leak detection spaced out at least 1 meter apart, which is a greater distance than 50 cm.Not found in available textThe available text does not contain any statement that sensors are spaced 'at least 1 meter apart' or any inter-sensor spacing distance; the only distances present are a 300-400 mm wheel diameter, a 50 mm tapped hole/gate valve, and pressures in kPa/MPa, none of which describe spacing between sensors. Paragraph ¶0091 is not visible in the provided excerpt (marked '[…description truncated]'), so the full specification should be checked. This is the quantitative limitation that the claim 17 rejection depends on entirely.
Claim 15 — sensors distributed in different locations about the pipeworkBond ¶0091 teaches sensors for leak detection spaced out at least 1 meter apart, which are different locations.Partially supportedThe available text supports the general concept via "It is possible to use a number of items of equipment each mounted at intervals along the length of the carrier" and the equipment "comprises a sensor, such as an acoustic sensor." However, the specific pin-cite ¶0091 and the specific quantitative teaching of spacing 'at least 1 meter apart' are not found in the available (truncated) text; the description excerpt ends with '[…description truncated]', so the full specification and ¶0091 should be checked for counsel to weigh.
4.

Element-by-Element Claim Chart

Claim 1 — §102 (Abbas)
Status glyphClaim elementStatusDisclosure / notesLocation
performing supervised training of a computer implemented leak prediction algorithm to predict leaks from pipework carrying a liquidAbbasDisclosedThe fully-grounded Abbas claims/summary plainly disclose supervised machine-learning training of a pipe-leak predictive model. This general limitation appears well supported for counsel.Abbas, claim 1 ('apply a supervised machine learning technique to generate a predictive model configured to determine a leak prediction of a pipe by training the predictive model'); Abbas, Brief Summary ('a predictive model can be generated using a supervised machine learning method for classifying outputs into a category (e.g., leak or no leak)')
that receives, as an input, training measurement data from sensors monitoring an environment in proximity to the pipeworkAbbasArguably disclosedCONTESTABLE for counsel. In the portions of Abbas actually in the record (claims + summary), the training input is described as pipe CHARACTERISTIC/attribute records (dimensions, material, age, location) supplied by an installer and stored in a pipe database — not environmental sensor measurement data. The examiner's support (¶¶0067, 0111) was not in the retrieved excerpt and should be pulled and read before relying on any distinction. Abbas is fully grounded, but the specific cited paragraphs remain unverified in this record.Abbas, claim 1 ('the first data items include characteristics of the respective pipes'); Abbas, Brief Summary ('information regarding the characteristics of various pipes (e.g., the dimensions of those pipes, the materials of those pipes, the age of those pipes, the locations of those pipes...)'); examiner cites Abbas ¶¶0067, 0111 (not present in the available excerpt)
provides, as an output, a prediction of whether a leak is likely to occur in futureAbbasDisclosedAbbas plainly outputs a future-leak prediction, including timing. Well supported in the fully-grounded text.Abbas, Brief Summary ('the system may be able to predict not only when a pipe will fail, but also at what time it will fail. For example, the system may predict a pipe will fail within 100 days'); Abbas, DETAILED DESCRIPTION ('used to accurately predict first time leaks in pipes')
wherein the supervised training comprises adjusting parameters of the computer implemented leak prediction algorithm to improve the accuracy of the predictionAbbasDisclosedTraining/validating a supervised model to improve accuracy is disclosed in the fully-grounded claims. The specific paragraphs the examiner cites for 'updating parameters over time' were not in the excerpt; the general concept is nonetheless supported.Abbas, claim 1 (training the predictive model; validating and generating a confusion matrix / true-positive & true-negative rates); examiner cites Abbas ¶¶0036, 0037, 0040 (not present in the available excerpt)
based on labels indicating which periods of the training measurement data correspond with one or more fault scenarios selected to cause leaks in futureAbbasArguably disclosedCONTESTABLE for counsel. In the fully-grounded text, Abbas labels each PIPE as leaked / not-leaked (a per-pipe known-leak label). The claim recites labels marking which PERIODS OF MEASUREMENT DATA correspond to 'fault scenarios selected to cause leaks in future' — a time-series/fault-scenario labeling concept that the available Abbas text does not describe. Whether the cited (unverified) ¶¶0037, 0040 supply this should be confirmed by pulling those paragraphs.Abbas, claim 1 ('known leaks associated with respective pipes'); Abbas, Brief Summary ('knowledge on whether those pipes leaked'); examiner cites Abbas ¶¶0037, 0040 (not present in the available excerpt)
Claim 2 — §103 (Abbas in view of Wang)
Status glyphClaim elementStatusDisclosure / notesLocation
wherein the training measurement data is obtained during controlled experiments in which the one or more fault scenarios are introduced to a system of pipework carrying a liquid so as to cause leaks at a period of time after the introduction of the one or more fault scenariosAbbas, WangArguably taughtVERIFY-FIRST posture only. Wang is an NPL reference whose text was never retrieved and is UNVERIFIABLE in this record; the examiner's characterization is taken as given and must not be contested here. Counsel should obtain and read Wang to confirm whether it in fact discloses introducing fault scenarios and collecting data 'at a period of time after' introduction before relying on any distinction. Separately note the §112(b) rejection: claim 2 is also rejected as indefinite for the relative term 'period of time' — a formal issue distinct from the prior-art teaching and one for counsel to address by amendment/clarification concept, not charted here.Wang — text NOT retrieved; examiner characterizes Wang (page 3, section 2; section 5) as an experimental setup introducing faults to a controlled pipeline platform and training a model on the resulting data
Claim 4 — §103 (Abbas in view of Mezghani)
Status glyphClaim elementStatusDisclosure / notesLocation
wherein the one or more fault scenarios comprises progressive decompression of a compression fittingAbbas, MezghaniArguably taughtCONTESTABLE for counsel. Mezghani (fully grounded) discloses a joint sealed with SOLDER or PVC CEMENT whose seal 'may deteriorate' over time — the examiner equates this seal deterioration with 'progressive decompression of a compression fitting.' The available Mezghani text describes a soldered/cemented seal, not a compression fitting, and describes seal deterioration rather than a controlled progressive decompression. Whether 'seal deterioration of a solder/PVC-cement joint' reads on 'progressive decompression of a compression fitting' is a point counsel may weigh.Mezghani, DETAILED DESCRIPTION ('an exemplary pipe joint is formed between a pipe P2 having a fitting F and a second pipe P1 seated in the fitting F, with solder or PVC cement forming a seal in the fitting. Over a period of time, the seal may deteriorate so that the pipes leak.'); examiner cites Mezghani ¶0016
Claim 6 — §103 (Abbas in view of Davis)
Status glyphClaim elementStatusDisclosure / notesLocation
wherein the sensors are placed within 30 centimeters (cm) of the pipeworkAbbas, DavisArguably taughtCONTESTABLE for counsel. Davis is fully grounded (claims retrieved) and discloses a sensor within a containment attached to/encapsulating the pipe, which supports close proximity generally. However, the specific numeric limitation 'within 30 cm' is not present in the available Davis text, and the exact paragraph the examiner relies on (¶0321) is in the truncated portion of the description. Counsel should pull ¶0321 / Figs. 1A-B to confirm the examiner's proximity read.Davis, claim 1 ('a sensor unit associated with the containment apparatus'; containment 'sealingly attachable to, and at least partially encapsulating, the fluid conduit system'); Davis, claim 7 ('the sensor is positioned adjacent the fluid accumulation area'); examiner cites Davis ¶0321 and Figs. 1A-B (¶0321 falls in the truncated portion of the description not in the available excerpt)
Claim 9 — §103 (Abbas in view of Reece)
Status glyphClaim elementStatusDisclosure / notesLocation
wherein the method further comprises training a second algorithm using measurement data that is classified as having a pattern by the first algorithmAbbas, ReeceArguably taughtCONTESTABLE for counsel. Reece is fully grounded but its stated framing is a leak / no-leak DETERMINATION, not a cascade in which a first algorithm classifies data 'as having a pattern' and a second algorithm is then trained on that classified data. The examiner's motivation ('predict what kind of leak will occur') is in tension with Reece's own framing (leak/no-leak, not leak-type classification). The specific ¶0064 (and the 'multiple classification models' assertion) is in the truncated Detailed Description; counsel should pull it to test whether Reece discloses the two-stage first-classifier/second-algorithm arrangement the claim recites. Depends on claim 8 (§102 over Abbas) — note claim 8's 'training a first algorithm to classify measurement data as including a pattern' is separately mapped to Abbas and not charted here.Reece, claim 1 (training a computer system on a first (normal) data set and a second (simulated-leak) data set to detect leaks); Reece, Summary/Field (deep learning models, neural networks, RNN, LSTM for leak / no-leak detection); examiner cites Reece ¶0064 (in the truncated Detailed Description, not in the available excerpt)
Claim 12 — §103 (Abbas in view of Mezghani)
Status glyphClaim elementStatusDisclosure / notesLocation
receiving measurement data from sensors monitoring an environment in proximity to the pipeworkAbbasArguably taughtSame contestable point as claim 1: the available Abbas text describes pipe-characteristic records rather than environmental sensor measurement data as the data source. Cited paragraphs unverified; pull ¶¶0067, 0111.examiner cites Abbas ¶¶0067, 0111 (not present in the available excerpt); Abbas, DETAILED DESCRIPTION (data sourced from a pipe database of pipe characteristic records supplied by a pipe installer)
providing the measurement data to a computerAbbasTaughtProviding data to a computer/system is broadly supported by the fully-grounded architecture description; the specific transmission paragraphs (¶¶0167-0168) were not in the excerpt.Abbas, FIG. 1 / DETAILED DESCRIPTION (pipe integration system stores pipe data in a pipe database used by the pipe leak prediction system); examiner cites Abbas ¶¶0167-0168 (not present in the available excerpt)
running a leak prediction algorithm on computer to process the measurement data and, responsive to an output from the leak prediction algorithmAbbasTaughtRunning the trained model to produce leak predictions is plainly disclosed.Abbas, claim 1 ('apply the predictive model to the pipeline dataset to determine leak predictions')
providing an alert in the event a leak is predictedMezghaniArguably taughtCONTESTABLE for counsel. Mezghani (fully grounded) plainly discloses sending an alert, but OA2 characterizes Mezghani as REACTIVE — it alerts when an actual leak is DETECTED (conductive liquid bridges the contacts), whereas the claim recites an alert 'in the event a leak is PREDICTED.' Whether an alert-on-detection teaches an alert-on-prediction is a point counsel may weigh, though the alert concept itself is disclosed.Mezghani, Abstract ('generating an alert signal when a liquid leak from the pipe joint or coupling is detected'); Mezghani, claim 1 ('activating the transmitter to transmit an alert signal to a remote receiver warning of a leak'); examiner cites Mezghani ¶0016
Claim 15 — §103 (Abbas in view of Mezghani and Bond)
Status glyphClaim elementStatusDisclosure / notesLocation
wherein the sensors are distributed in different locations about the pipeworkAbbas, Mezghani, BondArguably taughtCONTESTABLE for counsel. Bond is fully grounded but OA2 characterizes it as directed to DEPLOYMENT MECHANICS (a piston/winch system for inserting equipment into pressurized water mains), not to leak-detection sensor architecture. The available text does support multiple sensors mounted at intervals along a carrier (different locations), but the specific '1 meter apart' spacing (¶0091) is not in the available excerpt. Counsel may weigh whether Bond is analogous art for sensor distribution in a leak-prediction system and should pull ¶0091 to confirm the spacing teaching. Depends on claim 12.Bond, DETAILED DESCRIPTION ('It is possible to use a number of items of equipment each mounted at intervals along the length of the carrier'; 'the equipment comprises a sensor, such as an acoustic sensor'); examiner cites Bond ¶0091 ('spaced out at least 1 meter apart')
Claim 18 — §102 (Abbas)
Status glyphClaim elementStatusDisclosure / notesLocation
a plurality of environmental sensors disposed in proximity to the pipeworkAbbasArguably disclosedSame contestable point as claims 1 and 12: the available Abbas text does not describe a plurality of environmental sensors disposed near the pipework; it describes pipe-attribute records. Cited paragraphs unverified.examiner cites Abbas ¶¶0067, 0111 (not present in the available excerpt); Abbas, DETAILED DESCRIPTION (data source is a pipe database of pipe-characteristic records, not described as environmental sensors)
a computer, receiving environmental data measured by the plurality of environmental sensors and configured with a leak prediction algorithm that has been trained to predict leaks based on experimental data obtained during fault scenarios that will cause a leak in futureAbbasArguably disclosedCONTESTABLE for counsel and a key point for the §102 rejection. This limitation is asserted over Abbas ALONE. The claim requires the model to be 'trained ... based on experimental data obtained during fault scenarios that will cause a leak in future.' The fully-grounded Abbas text describes training on historical known-leak pipe records — not experimental fault-scenario data. This 'experimental / fault-scenario' training-data concept is what the examiner supplied via Wang/Reece for other claims, yet claim 18 relies on Abbas alone. Counsel should weigh whether Abbas anticipates this element. OMISSION NOTE (cap of 8 charts reached): rejected claims 3, 5, 7, 8, 10, 11, 13, 14, 16, 17, 19, 20 are not separately charted. Of note for counsel — claim 5/16/19 rely on Kamkalow, which is UNVERIFIABLE (WO number did not resolve; text never retrieved), so those sensor-type teachings must be verified before relying on any distinction; claims 3, 8, 10 (§102 over Abbas) and 20 (§102 over Abbas) largely track the claim 1 limitations charted above; claims 11 and 13 carry formal §112 rejections (antecedent-basis for 'second algorithm' in 11; improper multiple-dependency/conjunctive form in 13) that are formal matters for counsel rather than teaching-chart items; claim 17 (Bond ≥50 cm) tracks the claim 15 Bond analysis.Abbas, claim 1 (train predictive model on 'characteristics of the respective pipes' and 'known leaks'); Abbas, Brief Summary ('knowledge on whether those pipes leaked'); examiner cites Abbas ¶¶0026, 0066, 0076 (not present in the available excerpt)

Elements not shown by the cited art (4)

  • Claim 1 — “training measurement data from sensors monitoring an environment in proximity to the pipework”: Only Abbas is asserted. In the portions of Abbas actually in the record (claims and Brief Summary), the training input is pipe CHARACTERISTIC records (dimensions, materials, age, location) drawn from a pipe database populated by a pipe installer — the available fully-grounded text does not describe environmental sensors or sensor measurement data as the training input. CAVEAT: the examiner's cited support (¶¶0067, 0111) was NOT in the retrieved excerpt; although Abbas is graded fully grounded, those specific paragraphs must be pulled and read before treating this element as absent. Presented as a prima-facie-failure CANDIDATE for counsel to test, not a conclusion.
  • Claim 1 — “labels indicating which periods of the training measurement data correspond with one or more fault scenarios selected to cause leaks in future”: Only Abbas is asserted. The fully-grounded Abbas text discloses per-pipe 'known leaks' labels (a given pipe leaked or did not), not labels marking which time-periods of measurement data correspond to deliberately-introduced fault scenarios. CAVEAT: examiner's ¶¶0037, 0040 not in the retrieved excerpt — pull to verify. Candidate for counsel.
  • Claim 18 — “a leak prediction algorithm that has been trained to predict leaks based on experimental data obtained during fault scenarios that will cause a leak in future”: Only Abbas is asserted for this element (§102). The fully-grounded Abbas text describes training on historical known-leak pipe records, not on 'experimental data obtained during fault scenarios.' The experimental/fault-scenario training concept is the very teaching the examiner drew from Wang (claim 2) and Reece (claim 9) for other claims, suggesting Abbas alone may not supply it. CAVEAT: examiner's cited ¶¶0026, 0066, 0076 were not in the retrieved excerpt and should be pulled. Candidate for counsel.
  • Claim 4 — “progressive decompression of a compression fitting”: Mezghani is fully grounded and is the reference relied on. The available Mezghani text discloses a joint sealed with SOLDER or PVC CEMENT whose seal 'may deteriorate' over time — not a compression fitting, and not a 'progressive decompression.' Whether solder/PVC-cement seal deterioration reads on 'progressive decompression of a compression fitting' is a genuine gap for counsel to weigh (examiner ¶0016 relied upon).
5.

Rejection Map

§112(b)Indefiniteness — claims 2

The term 'period of time' in claim 2 is a relative term that renders the claim indefinite. The term is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope. The examiner states it is not clear as to when the leak can appear.

§112(b)Indefiniteness — claims 11

Claim 11 recites 'the first algorithm and second algorithm' in the first line but there is insufficient antecedent basis for this limitation in the claim. Claim 11 depends on claim 8, which introduces only 'a first algorithm'; the 'second algorithm' is introduced in claim 9, which is not in claim 11's dependency chain.

OtherRejection — claims 13

Rejected under 35 U.S.C. 112(e) as an improper multiple dependent claim. Claim 13 refers to two claims in the conjunctive (claim 12 AND claim 1) rather than the alternative (claim 12 OR claim 1). The examiner states the claim cannot be further examined on its merits based on the improper claim dependency.

§102Anticipation — claims 1, 3, 8, 10, 18, 20

AbbasUS 20180300639 A1

Abbas anticipates under 102(a)(2). For claim 1: Abbas ¶¶0067, 0111 teach training a random forest algorithm for leak prediction using sensor data; ¶¶0026, 0066, 0076 teach prediction of future leaks; ¶¶0036, 0037, 0040 teach updating model parameters with new data to improve prediction; ¶¶0037, 0040 teach data with information about pipes that have leaked and causes of leaks for training. Claim 3: ¶0032 teaches predicting leaks before they happen or become significant. Claim 8: ¶0037 teaches classifying patterns related to leak detection using measurement data. Claim 10: ¶¶0030-0031 teach the algorithm can be a neural network. Claims 18 and 20 are mapped to the same Abbas paragraphs as claim 1.

§103Obviousness — claims 2MPEP §2143(G)

AbbasUS 20180300639 A1NPLWang

Abbas teaches claim 1 but not obtaining training data during controlled experiments with introduced fault scenarios. Wang (page 3, section 2) teaches an experimental setup where faults are introduced to a controlled experimental pipeline platform and data from faults are used to train a prediction model (section 5). Motivation: to train the algorithm on data for leaks that might not have a lot of data associated with them.

§103Obviousness — claims 4, 12, 14MPEP §2143(G)

AbbasUS 20180300639 A1MezghaniUS 20110227721 A1

For claim 4: Abbas does not teach progressive decompression of a compression fitting. Mezghani ¶0016 teaches detecting pipe leaks at the point of a fitting compressed between two pipes caused by seal deterioration (i.e. progressive decompression). Motivation: to train the algorithm to predict leaks caused by joining of two pipes. For claim 12: Abbas teaches receiving measurement data (¶¶0067, 0111), providing to computer (¶¶0167-0168), and running leak prediction (¶¶0026, 0066, 0076). Abbas does not teach providing an alert; Mezghani ¶0016 teaches sending an alert when a leak is detected. Motivation: to alert users when a leak is about to or has occurred. Claim 14: Abbas ¶¶0167-0168 teaches transmitting measurement data to the computer system.

§103Obviousness — claims 5, 16, 19MPEP §2143(G)

AbbasUS 20180300639 A1KamkalowWO9941580A1

Abbas does not teach sensors comprising humidity, temperature, and atmospheric pressure sensors. Kamkalow (page 3, ¶4) teaches a pipe leak detection system utilizing temperature, humidity and pressure sensors. Motivation: to detect and track changes in common factors that lead to leaking of pipes.

§103Obviousness — claims 6, 7MPEP §2143(G)

AbbasUS 20180300639 A1DavisUS 20160284193 A1

For claim 6: Abbas does not teach sensors placed within 30 cm of pipework. Davis ¶0321 teaches a sensor affixed to a container attached to the pipe, with fig. 1A-B showing the sensor touching the pipe. For claim 7: Abbas does not teach sensors placed in an enclosed cavity with the pipework. Davis ¶0321 teaches a sensor affixed to an enclosed container (i.e. cavity) attached to the pipe. Motivation: to more accurately detect leaks with less interference from outside factors on the sensors.

§103Obviousness — claims 9, 11MPEP §2143(G)

AbbasUS 20180300639 A1ReeceUS 20210216852 A1

For claim 9: Abbas does not teach training a second algorithm using data classified by the first algorithm. Reece ¶0064 teaches use of multiple classification models to predict and classify leaks based on measurement data. For claim 11: Reece ¶¶0013, 0064 teaches multiple classification models which can be neural networks. Motivation: to not only predict if a leak will occur but also predict what kind of leak will occur.

§103Obviousness — claims 15MPEP §2143(G)

AbbasUS 20180300639 A1MezghaniUS 20110227721 A1BondUS 20020148294 A1

The combination of Abbas and Mezghani does not teach sensors distributed in different locations about the pipework. Bond ¶0091 teaches sensors for leak detection spaced at least 1 meter apart (different locations). Motivation: to more accurately collect data on the length of the pipe system and detect leaks more accurately at different locations.

§103Obviousness — claims 17MPEP §2143(G)

AbbasUS 20180300639 A1KamkalowWO9941580A1BondUS 20020148294 A1

The combination of Abbas and Kamkalow does not teach sensor units spaced at least 50 cm apart. Bond ¶0091 teaches sensors for leak detection spaced at least 1 meter apart, which is greater than 50 cm. Motivation: to more accurately collect data on the length of the pipe system and detect leaks more accurately at different locations.

References Cited

6.

Record & Grounding

Grounding Summary

Note

How each cited reference was grounded. A reference the analysis could only read through the office action’s characterization is flagged — its findings are limited to what the examiner said, not the reference itself.

NOVEL AUTONOMOUS ARTIFICIALLY INTELLIGENT SYSTEM TO PREDICT PIPE LEAKSUS20180300639A1
Claim text retrieved
Leak detection system for pipesUS20110227721A1
Claim text retrieved
KamkalowWO9941580A1
Not retrieved — analysis limited to the OA's characterizationthis number did not resolve to a document — verify it
FLUID LEAK DETECTION METHODS, SYSTEMS AND APPARATUSUS20160284193A1
Claim text retrieved
LEAK DETECTION WITH ARTIFICIAL INTELLIGENCEUS20210216852A1
Claim text retrieved
Deployment of equipment into fluid containers and conduitsUS20020148294A1
Claim text retrieved

Data Egress Log

Note

Your uploads stay in-boundary. External retrieval was limited to public patent-number lookups: 5 fetches. No claim text, no client material left the environment.

Documents processed
  • 51b703a4-4cb6-49cb-be73-35dace10f0c9.pdfoffice action
  • 0ef4e235-4f71-482c-8e9c-7438fa214ced.pdfclaims
Processed in-boundary — never transmitted externally.

Obviousness Framework

Field of endeavor
Computer-implemented machine-learning systems and methods for predicting leaks from pipework carrying a liquid (e.g., water mains), including training a supervised leak-prediction algorithm on labeled measurement data from sensors monitoring the environment in proximity to the pipework.
PHOSITA
For argument purposes (a proposed construction, not a factual finding, for counsel to adopt or adjust): a person with a bachelor's-level education in an engineering or computer/data-science discipline together with a few years of practical experience in either (a) machine-learning/predictive-modeling for classification and time-series prediction, or (b) fluid-conveyance / water-distribution pipe monitoring and leak detection instrumentation — and who could consult with a counterpart from the other of those two disciplines. This person would be familiar with supervised training, model parameter adjustment, sensor deployment, and conventional leak-detection instrumentation.A construction for argument — not asserted as fact.
ReferenceAnalogous artRationale
AbbasAnalogousSame field of endeavor: a computer-implemented machine-learning system that trains a supervised predictive model to predict pipe leaks. Directly in the claimed field. Note for counsel: OA2 flags that the available Abbas excerpt describes training on pipe characteristic/attribute records (dimensions, materials, age, location, contents) rather than on environmental-sensor measurement data, and that the specific paragraphs the examiner cites (e.g., ¶¶0067, 0111, 0167-0168) are not in the available excerpt — analogous-art status is not in dispute, but the content mapping to the claimed sensor-based training input is.
WangAnalogousAs characterized in the office action, Wang concerns water-pipeline leak study using acoustic signal analysis and artificial-neural-network prediction — same field (ML-based leak prediction for water pipes). Flag for counsel: the actual Wang text is NOT in the provided record (only the examiner's characterization is available), so the analogous-art conclusion rests on the OA's description and should be verified against the reference itself.
MezghaniContestableSame broad field of leak detection from pipes, so prong (1) is arguable. However, per OA2, Mezghani is a reactive hardware system (spaced contacts bridged by conductive liquid closing a transmitter circuit) that detects an actual leak — it involves no machine learning and no prediction. Whether it is 'reasonably pertinent to the particular problem' the inventor faced (training a predictive ML algorithm) is contestable; its pertinence is limited to leak detection generally and to the fitting/seal-deterioration mechanism.
KamkalowAnalogousAs characterized in the OA, Kamkalow is a pipe leak-detection system using temperature, humidity, and pressure sensors — same field of endeavor (leak detection from pipes) and pertinent to the environmental-sensing aspect of the inventor's problem. Flag for counsel: the actual Kamkalow (WO9941580A1) text is NOT in the provided record, so this rests on the examiner's characterization and must be verified.
DavisAnalogousSame field of endeavor: fluid leak detection from pipes/conduits using a sensor unit associated with a containment apparatus and an alarm. Flag for counsel: OA2 states the available Davis text is truncated well before ¶0321 (the paragraph the examiner relies on for sensor placement within 30 cm / in an enclosed cavity), so the specific placement teaching is not verifiable in the record.
ReeceAnalogousSame field of endeavor: an AI/deep-learning computer-implemented method of detecting leaks in a pipeline, trained on normal-operation data and simulated-leak data. Squarely in the claimed field. Flag for counsel: the Detailed Description is truncated, so ¶0064 (cited by the examiner) is not verifiable in the record, and OA2 notes Reece's framing is a leak/no-leak determination, not leak-type classification.
BondContestablePer OA2, Bond is directed to the MECHANICS of deploying equipment (which may be a sensor or transmitter) into pressurized fluid containers/conduits (e.g., water mains) using a piston/fluid-housing and a winch/carrier — not to leak detection algorithms or to sensor spacing for leak detection. Prong (1) same-field is doubtful (deployment apparatus vs. leak-prediction). Prong (2) reasonable-pertinence to the inventor's problem (placement/spacing of environmental sensors about pipework for leak prediction) is contestable, since Bond's mention of equipment 'mounted at intervals along the carrier' arises in a deployment-into-a-conduit context. Flag for counsel: the specific ¶0091 '1 meter apart' spacing the examiner relies on does not appear in the provided Bond excerpt and should be pulled and verified. Counsel may wish to develop a non-analogous-art position (MPEP § 2141.01(a)).

Claim 2 rejected under §103 over Abbas in view of Wang; Wang asserted to teach obtaining training data during controlled experiments in which fault scenarios are introduced to a pipework system to cause leaks after a period of time.

Abbas + Wang

Motivation asserted To be able to train the algorithm on data for leaks that might not have a lot of data associated with them.

  • Othermoderate

    The Wang reference text is not in the provided record — only the examiner's characterization is available. Counsel cannot verify from the record that Wang actually discloses introducing fault scenarios to cause leaks 'at a period of time after the introduction,' as claim 2 recites. The reference should be pulled and the mapping verified before the combination can be assessed on its merits.

  • Othermoderate

    Per OA2, the available Abbas text trains its model on pipe characteristic/attribute records rather than on environmental-sensor measurement data. Claim 2 depends from claim 1's 'training measurement data from sensors monitoring an environment,' so the combination presupposes an Abbas teaching (sensor-based training input) that the record excerpt does not clearly show; this is a base-reference content gap for counsel to weigh against the prima facie case.

  • Conclusory motivationweak

    The asserted reason to combine ('to train on data for leaks that might not have a lot of data') is a general benefit statement; whether it rests on an articulated rational underpinning tied to the references' actual teachings (MPEP §§ 2143, 2143.01) is for counsel to test, particularly given that claim 2 is independently rejected under §112(b) for the same 'period of time' term.

Claims 4, 12, and 14 rejected under §103 over Abbas in view of Mezghani. Mezghani asserted to teach (claim 4) leak at a compressed fitting from seal deterioration read as 'progressive decompression,' and (claim 12) sending an alert when a leak is detected.

Abbas + Mezghani

Motivation asserted For claim 4, to train the algorithm to predict leaks caused by the joining of two pipes together and not just single pipes; for claim 12, to alert users when a leak is about to or has already occurred in order to prevent or fix the leak.

  • Othermoderate

    Mezghani, per OA2, describes natural, gradual seal deterioration ('Over a period of time, the seal may deteriorate so that the pipes leak') — it does not describe deliberately introducing 'progressive decompression of a compression fitting' as a controlled fault scenario. Whether Mezghani's passive seal-deterioration teaching reads on the recited fault scenario is a claim-mapping question for counsel to test.

  • Destroys principle of operationmoderate

    For claim 12, Mezghani's alert is triggered by an ACTUAL leak (conductive liquid bridging the contacts to close a circuit), i.e., a reactive detection system, whereas claim 12/Abbas concern providing an alert 'in the event a leak is PREDICTED.' Grafting Mezghani's actual-leak-triggered transmitter onto a predictive system arguably does not supply the predictive-alert limitation and may reflect a mismatch in operating principle that counsel can develop (MPEP § 2143.01).

  • Conclusory motivationweak

    The motivation statements are general benefits (predicting fitting leaks; alerting users). Counsel may test whether the examiner articulated a rational underpinning for why a PHOSITA would combine a passive electrical-contact leak alarm with Abbas's ML predictive model rather than merely recite a desirable outcome (MPEP § 2143).

Claims 5, 16, and 19 rejected under §103 over Abbas in view of Kamkalow. Kamkalow asserted to teach a leak-detection system using temperature, humidity, and pressure sensors.

Abbas + Kamkalow

Motivation asserted To be able to detect and track changes in common factors that lead to the leaking of pipes in pipe systems.

  • Othermoderate

    The Kamkalow (WO9941580A1) text is not in the provided record; only the examiner's characterization (page 3, ¶4) is available. Counsel cannot verify from the record that Kamkalow discloses the specific temperature/humidity/atmospheric-pressure sensor combination, so the reference should be pulled and verified before assessing the combination.

  • Othermoderate

    Per OA2, Abbas's available text trains on pipe characteristic records, not on environmental sensor data. Adding environmental (temperature/humidity/pressure) sensors as the training input arguably alters Abbas's data model rather than being a simple substitution of one equivalent element for another; counsel can weigh whether the examiner's rationale (MPEP § 2143(B)/(C)) fits given this gap.

  • Conclusory motivationweak

    The stated motivation ('detect and track changes in common factors that lead to leaking') is a generalized benefit; counsel may test whether it is supported by an articulated line of reasoning tied to the references' actual teachings.

Claims 6 and 7 rejected under §103 over Abbas in view of Davis. Davis asserted (¶0321, figs. 1A-B) to teach a sensor within 30 cm of the pipe (claim 6) and a sensor in an enclosed cavity with the pipework (claim 7).

Abbas + Davis

Motivation asserted To be able to more accurately detect leaks with less interference from outside factors on the sensors.

  • Othermoderate

    Per OA2, the available Davis text is truncated well before ¶0321 — the very paragraph the examiner relies on for the '30 cm' proximity and 'enclosed cavity' teachings. Counsel cannot verify these specific mappings from the record; the full Davis text should be pulled before the combination is assessed.

  • Conclusory motivationweak

    The motivation ('more accurately detect leaks with less interference') is a general benefit; whether the examiner supplied a rational underpinning for a PHOSITA to adopt Davis's containment/sensor placement in Abbas's predictive ML system (MPEP § 2143) is for counsel to test, especially given Davis is a reactive containment-and-alarm apparatus rather than a training methodology.

Claims 9 and 11 rejected under §103 over Abbas in view of Reece. Reece (¶0064; ¶¶0013, 0064) asserted to teach multiple classification models (which can be neural networks) used to predict and classify leaks, mapped to training a second algorithm on data classified by a first algorithm.

Abbas + Reece

Motivation asserted To be able to not only predict if a leak will occur but also predict what kind of leak will occur.

  • Otherstrong

    The asserted motivation rests on Reece enabling prediction of 'what kind of leak' will occur, but OA2 states Reece's actual framing is a leak/no-leak determination and that its 'stated framing is detecting/alarming on leaks ... not classifying leak types.' If the reference reality check is correct, the examiner's stated reason to combine relies on a capability Reece does not disclose — a mismatch between the motivation and the reference's teaching that counsel can develop.

  • Othermoderate

    The examiner cites Reece ¶0064 (and ¶0013) for a two-stage/multiple-classifier arrangement, but the Reece Detailed Description is truncated in the record and ¶0064 is not present in the available excerpt. Claim 9 specifically requires a SECOND algorithm trained USING data CLASSIFIED by the FIRST algorithm; whether Reece discloses that sequential classify-then-train relationship (as opposed to training on normal vs. simulated-leak data) cannot be verified from the record and should be pulled.

  • Hindsight reconstructionmoderate

    Because the mapped two-stage classifier relationship (claim 9) is not verifiable in the record and the motivation appears tied to a leak-type-classification capability the reality check says Reece lacks, counsel may argue the combination is assembled with the claim as a template rather than from an articulated teaching (MPEP § 2143.01). Independently, claim 11's §112(b) antecedent-basis rejection ('the first algorithm and second algorithm') bears on whether claim 11 can be reached on the merits.

Claim 15 rejected under §103 over Abbas in view of Mezghani, further in view of Bond. Bond (¶0091) asserted to teach sensors spaced at least 1 meter apart, mapped to sensors distributed in different locations about the pipework.

Abbas + Mezghani + Bond

Motivation asserted To be able to more accurately collect data on the length of the pipe system and detect leaks more accurately at different locations in the system.

  • Otherstrong

    Bond is directed to the mechanics of deploying equipment into pressurized fluid conduits (per OA2), not to leak detection or sensor placement for leak prediction. Counsel may develop a non-analogous-art position (MPEP § 2141.01(a)): Bond is arguably outside the claimed field of endeavor and its pertinence to the inventor's problem of distributing environmental sensors about pipework for leak prediction is contestable. A non-analogous reference cannot support a §103 rejection.

  • Othermoderate

    The specific ¶0091 teaching of sensors 'spaced at least 1 meter apart' does not appear in the provided Bond excerpt; the available text refers only to equipment 'mounted at intervals along the carrier' in a deployment context. Counsel cannot verify the cited spacing figure from the record and should pull the full reference.

  • Hindsight reconstructionmoderate

    Reaching to a deployment-apparatus reference for a sensor-spacing figure to satisfy 'distributed in different locations' — when Bond's context is inserting equipment along a carrier inside a conduit rather than placing environmental sensors about pipework — invites the argument that the combination is reconstructed with knowledge of the claim (MPEP § 2143.01), particularly since the motivation is a generalized benefit statement.

Claim 17 rejected under §103 over Abbas in view of Kamkalow, further in view of Bond. Bond (¶0091) asserted to teach sensors spaced at least 1 meter apart, which is greater than the claimed 50 cm.

Abbas + Kamkalow + Bond

Motivation asserted To be able to more accurately collect data on the length of the pipe system and detect leaks more accurately at different locations in the system.

  • Otherstrong

    As with claim 15, Bond is directed to equipment-deployment mechanics rather than leak detection or environmental-sensor spacing (per OA2). Counsel may develop a non-analogous-art challenge (MPEP § 2141.01(a)); if Bond is neither in the same field of endeavor nor reasonably pertinent to the inventor's problem, it cannot support the §103 rejection.

  • Othermoderate

    The cited ¶0091 spacing ('at least 1 meter apart') is not present in the provided Bond excerpt, and the Kamkalow (WO9941580A1) text is not in the record at all. Two of the three references in this stacked rejection thus have their key mapped teachings unverifiable from the record and should be pulled and confirmed.

  • Conclusory motivationweak

    The motivation is the same generalized 'more accurate data/detection' statement reused across rejections; counsel may test whether an articulated, record-grounded rationale supports importing Bond's deployment spacing into a combined Abbas/Kamkalow environmental-sensing system (MPEP § 2143).

This site uses cookies to improve your experience.