US 10274338

I'll run targeted searches for this patent number and any associated litigation, then reconcile with the authoritative full text provided.

Let me run a few more targeted searches for litigation/IPR activity on this specific patent.

Summary: US 10,274,338 B2

Important sourcing note up front: Everything below is grounded in the full patent text you supplied (fetched from Google Patents) plus my searches. My searches returned no litigation, PTAB/IPR, or CAFC 2026 docket activity involving US 10,274,338. I can only say I did not find such activity with the queries I ran — this is not proof that none exists. I did not have direct access to a USPTO PatentCenter or PACER docket for this number, and public assignment records were the closest thing to authoritative corroboration I could retrieve. Treat the absence of litigation findings as "not found," not "confirmed none."

Bibliographic data

Prosecution nuance worth flagging: The granted independent claims differ from the as‑published pre‑grant application. The published version (US 2018/0162307 A1) recited "executing a context modifying action…", while the granted claims recite "autonomously executing by the vehicle a context modifying action… wherein the context modifying action is a personalized action performed by the vehicle to take over and autonomously execute…". The granted claim 8 preamble is "A vehicle system" (published was "A system"), and granted claim 15 recites a "non‑transitory computer readable storage medium" (published said "computer readable storage"). So the issued claims were narrowed toward autonomous vehicle‑executed action. (The patent also carries a Certificate of Correction dated April 27, 2021, and a corrective assignment in Feb. 2021 fixing an inventor‑name typographical error.)

Abstract (as issued)

A method for improving risk situations for vehicle occupants which includes: configuring a set of circumstances; defining values for each circumstance where each value has a rate; collecting context information; collecting real‑time sensor measurements pertaining to a vehicle, driver, and occupants; retrieving risk patterns from a risk pattern database; matching the sensor measurements to the risk patterns to find a matching pattern having a risk similarity value; contextualizing the matching risk pattern by increasing the risk similarity value to result in a personalized risk value; comparing the personalized risk value to a threshold; and executing a context modifying action to lower the personalized risk value below a predefined threshold when the personalized risk value exceeds the predefined threshold. (Note: the abstract omits the "autonomously executing by the vehicle" language that appears in the granted independent claims.)

Plain-language overview of the independent claims

Claim 1 — Computer-implemented method

1. Configure a set of "circumstances," each being a condition of time, place, or manner that accompanies/influences an event or person.
2. Define values for each circumstance, each value having a "rate" (a weighting).
3. Collect context information for the circumstances/values/rates.
4. Collect, via vehicle sensors, real‑time measurements about the vehicle, driver, and occupants.
5. Retrieve risk patterns from a risk pattern database.
6. Match the sensor measurements to the risk patterns to find a matching risk pattern with a "risk similarity value."
7. "Contextualize" the matching pattern by increasing the risk similarity value to yield a "personalized risk value."
8. Compare the personalized risk value to a threshold.
9. If it exceeds a predefined threshold, the vehicle autonomously executes a context modifying action — a personalized action where the vehicle takes over and performs it with respect to the occupants or vehicle — to bring the personalized risk value back below the threshold.

Claim 8 — Vehicle system
Same nine functional steps, but framed as a system: a computer readable storage medium storing instructions plus a processor executing them to perform the identical functions. Narrowed from the published "system" to a "vehicle system."

Claim 15 — Computer program product
Same nine steps, embodied as program instructions on a non‑transitory computer readable storage medium.

Representative dependent claims (mirrored across the three independents):

• Repeating the process from the start when the personalized risk value does not exceed the threshold (claims 2/9/16), or when no matching risk pattern is found (3/10/17).
• Context information includes driver identification/profile, driver mood, time of day, weather, GPS location, and route info (4/11/18).
• Continuous real‑time sensor collection (5/12).
• A training process: predefine default context/rates and default risk patterns, then adjust the rates using collected context and sensor data (6/13/19).
• The context modifying action is a "context modifier" that changes environmental variables to reduce risk (7/14/20).

Supplementary observations from the file

• Prior art cited on the face (21 references), including US 8,892,451 (Progressive), US 6,335,689 (Fuji Jukogyo), US 6,925,425 (Motorola), US 9,047,170 (Naboulsi), US 8,508,351 (Bosch), US 8,849,501 / US 8,508,353 (Lytx/Drivecam), US 2017/0076395 (Inrix), US 2017/0089710 (Allstate), and US 9,791,864 (Ford). Six non‑patent citations, e.g., Remboski et al. "Driver Performance Improvement Through the Driver Advocate" (SAE, 2000) and Rendon‑Velez et al. "Progress With Situation Assessment and Risk Prediction in Advanced Driver Assistance Systems: A Survey" (2009).
• Forward citations (citing this patent) referenced in the record include US 2021/0319238 A1 and US 12,198,448 B2 (both Caroline Bono), and EP 4,105,842 A1 / US 12,394,320 B2 (Honeywell). Also a Kyndryl‑owned sibling, US 11,299,174 B2 ("Dual‑test operator assessment").
• Worked examples in the spec include a "person left alone in a vehicle" detection (five‑row sensor log; actions: open window 1 cm, send SMS alert), and a high‑speed‑at‑night‑in‑rain scenario where a 60% risk similarity value is increased by a 25% stress factor to a 75% personalized risk value against a 70% threshold, triggering "turn on the radio with learned driver preferences for soft music." These examples are useful context for how broadly the claim language maps to the disclosed automation.

Uncertainty / caveats

• Litigation: No CAFC 2026, PTAB, or district court matter asserting US 10,274,338 was found. I cannot rule out proceedings that are not indexed by the search tools I used.
• Term: The 2037‑05‑23 adjusted expiration is what Google Patents lists; the base 20‑year term from the Dec. 11, 2016 filing would be Dec. 11, 2036, so the extra time appears attributable to patent term adjustment. I did not independently verify the PTA calculation.
• Inventor name: The record shows a typo correction — the originally recorded "Ivonne R. Cuervo" is the same person as "Ivonne Rocio Cuervo Fajardo" per the corrective assignment. I report both spellings as they appear rather than auto‑correcting.
• The "Definitions" and "Classifications" text on the Google Patents page appears partly machine‑assembled and somewhat duplicative; where it conflicted, I relied on the authoritative claim and description text.

I'll search for litigation involving US patent 10274338 across patent litigation sources.

Let me search more specific litigation sources.

Let me run additional targeted searches to be thorough.

Litigation Search Report — U.S. Patent No. 10,274,338

Bottom Line

I found no litigation (district court, ITC, CAFC, or PTAB) involving U.S. Patent No. 10,274,338. Based on my searches, there is no known infringement suit or administrative challenge directed to this specific patent.

What I Searched and What I Found

Every hit containing the string "743" or "338" in my results belonged to different patents (e.g., U.S. Pat. 8,252,838, the free-flow packaging '743 patent, 10x Genomics '224/'007 patents, Stratasys patents, Lexmark '165/'711). None matched the digits 10,274,338 literally.

Important Caveats (Read Before Relying on This)

1. Absence of evidence is not evidence of absence. My search coverage here was limited (a handful of web searches; I did not successfully query Unified Patents' case list, Docket Navigator, PACER, or the CAFC docket directly — the PTAB and Kyndryl-specific queries were cut off by a search-step limit). A quiet, non-practicing-entity patent like this one often has no litigation, but I cannot certify that from these results alone.

2. This patent is an unlikely litigation candidate on its face. US 10,274,338 is a software/analytics method patent ("Risk situations for vehicle occupants based on data provided by vehicle sensors and contextual information"), originally assigned to IBM and now held by Kyndryl, Inc. (assignment effective 2021-09-30). IBM/Kyndryl have historically asserted only select patents, and this one is not among the frequently litigated automotive patents.

3. The "Cited By" and "Families Citing" data in the patent record are not litigation. Those entries (e.g., Honeywell EP4105842A1/US12394320B2; Caroline BONO US20210319238A1/US12198448B2; Micron US11091166B1/US11494865B2) are later publications that cite this patent as prior art — they are patent citations, not lawsuits involving the '338 patent.

4. No reassignment-to-a-litigation-entity signal. The chain of title (IBM → Kyndryl) shows no transfer to a known patent-assertion entity (PAE), which is the common trigger for NPE litigation.

Recommendation

To definitively confirm "no litigation," the following primary sources should be checked directly (I was unable to fully complete these):

• Unified Patents litigation portal: https://portal.unifiedpatents.com/litigation/caselist
• USPTO PTAB (PATTS/PTAB API): https://ptacts.uspto.gov — search Patent Number 10274338 for any IPR/PGR/CBM
• PACER / CourtListener RECAP: search "10,274,338"
• CAFC docket (via PACER or the Federal Circuit's own docket)
• RPX Insight / Docket Navigator (subscription databases)

If you want, I can re-run these specific queries now that the tool is available again, or I can provide a formatted declaration stating that a reasonable search identified no litigation as of the current date.

PTAB proceedings on file — US 10,274,338 B2

Bottom line: there are no AIA trial proceedings on this patent. The structured "PTAB proceedings on file" block (USPTO Open Data Portal ingest) returns an empty set for US 10,274,338, and four independent web searches for IPR/PGR/CBM activity on this patent number, its application number (15/375,136), and its pre‑grant publication (US 2018/0162307 A1) surfaced nothing. I am not going to manufacture a proceeding number to fill the template — the correct answer is that this patent has never been before the PTAB.

Proceedings overview

Total AIA trial proceedings on file: 0 — 0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution‑denied. Consequently there is no FWD, no Certificate under 35 U.S.C. § 318(b), no joinder, and no CAFC appeal to report, because there is no trial to have appealed from.

The defensive posture this gives a defendant is the unusual one: all 20 claims are untested and, critically, unbeaten. Claim 1 is not "dead"; there is no IPR‑based path to a quick, cheap kill, and no estoppel running in your favor against Kyndryl. Conversely, there is also no hardened, PTAB‑blessed claim set that a plaintiff can wave at the court — the claims have drawn zero adversarial scrutiny post‑grant, which is itself informative about how little this patent has been asserted. Treat the empty PTAB docket as "no free weapons, no wounds either."

There are no proceedings to enumerate

For completeness, the template categories map as follows:

Verification paths I used and would point a defendant to:

• PTAB End‑to‑End (E2E) / PTAB Center party and patent‑number search: https://ptacts.uspto.gov/ptacts/
• USPTO Open Data Portal AIA trial APIs (the canonical source for the structured block above)
• USPTO PTAB Decisions and AIA trial statistics: https://www.uspto.gov/patents/ptab/statistics
• CourtListener / CAFC docket search for any appeal naming the patent: https://www.courtlistener.com/?type=o&q=%2210%2C274%2C338%22

Strategic summary

Claim status across the patent. All 20 claims remain in force and none has been canceled, disclaimed, or amended in any PTAB proceeding. Independent claims 1 (method), 8 (vehicle system), and 15 (computer program product) are all untested; dependent claims 2–7, 9–14, and 16–20 are likewise untested. The only narrowing on the face of the record is the prosecution‑stage narrowing already identified in the earlier section — the granted independents add "autonomously executing by the vehicle" and "personalized action performed by the vehicle to take over and autonomously execute," which is a real limitation your non‑infringement story can attack (the vehicle must autonomously take over; a mere driver‑facing alert arguably does not read on claim 1). That narrowing was obtained in examination, not in an IPR, and is not backed by any PTAB or CAFC construction.

Estoppel landscape. With no IPR and no FWD under § 318(a), no § 315(e)(1) or § 315(e)(2) estoppel has attached to anyone. No petitioner is barred, no privy is barred, and no district‑court invalidity ground is foreclosed. This cuts both ways: it means you are free to file an IPR on any § 102/§ 103 ground built on patents and printed publications, and it means no competitor has already burned the good art in a way that helps you. Note the related practical constraints for a newly‑sued defendant: the § 315(b) one‑year clock runs from service of a complaint alleging infringement, and § 315(a)(1) bars an IPR if you filed a DJ action of invalidity first. Also weigh § 325(d) — because this patent was examined in a relatively crowded art area (the 21 references of record plus six NPL citations, including Remboski's Driver Advocate and the Rendon‑Velez ADAS survey), a petition relying on art or arguments the examiner already considered risks discretionary denial.

Pattern signals. There are no pattern signals to read: no repeat petitioner, no serial petitions, no defensive aggregator (no Unified Patents or RPX involvement surfaced), no patent‑owner appeal activity, no settled‑then‑refiled campaign. The forward citations in the record are all ordinary later patents (Caroline Bono; Honeywell; a Kyndryl sibling, US 11,299,174 "Dual‑test operator assessment"), not litigation artifacts. The combination of (a) a 2016‑filed, 2019‑granted, currently‑active patent and (b) zero PTAB petitions more than seven years post‑grant strongly suggests the patent has rarely or never been asserted in a campaign-style manner, which is the usual trigger for IPR filings. I found no district‑court assertion of this patent either — but treat that as "not found," not "confirmed none," since I did not have PACER/Docket Navigator access for this number.

Recommended next steps

If you are a defendant and want a PTAB kill:

1. You still can file. There is no estoppel and no prior FWD. Because the patent issued 2019‑04‑30, PGR is time‑barred and CBM is unavailable, so IPR (35 U.S.C. §§ 102/103 on patents and printed publications) is your only AIA vehicle. Confirm your § 315(b) window relative to the date you were served.
2. Attack the "autonomous vehicle take‑over" limitation. The granted claim language is the narrowest thing in the file: "autonomously executing by the vehicle a context modifying action… wherein the context modifying action is a personalized action performed by the vehicle to take over and autonomously execute the personalized action." Art showing a driver‑alert‑only or cloud‑side risk score, without the vehicle autonomously taking over the occupants' environment, is a natural § 102/§ 103 theory.
3. Do not expect help from an existing FWD — there isn't one. The absence of PTAB activity means the whole invalidity case is yours to build from scratch. Budget for it accordingly; there is no "adopt the IPR record" shortcut.
4. Consider the prosecution record as a § 325(d) risk and an opportunity. The 21 cited references and 6 NPL items give you a ready‑made map of what the examiner saw — useful for distinguishing, but also the reason to lead with art the examiner did not consider.

If you are advising on or monitoring this patent: there is nothing to monitor at the PTAB. Monitor PTAB E2E (https://ptacts.uspto.gov/ptacts/) for any newly docketed petition, and CourtListener/PACER for a first infringement suit, since a first assertion is what would normally precipitate the first IPR.

If there is in fact a proceeding I could not surface: the ODP ingest and my searches could miss a very recently filed petition (ODP lag is real, and PTAB E2E petition pages are frequently not indexed by general web search). If you have an E2E hit, give me the proceeding number and I will pull the institution decision, panel, and FWD at claim‑level granularity — but I will not guess a number.

Caveats

• Source of truth. The "no proceedings" conclusion rests on the structured USPTO ODP block you supplied plus negative search results. It is a "not found" conclusion for the searches, and a "no proceedings on file" conclusion per ODP as of the ingest date.
• False‑positive caution. The string "10274338" also appears as a DOI suffix in an unrelated 2024 sports‑science article; patent‑number searches on bare digit strings will pick up noise like that. I filtered for PTAB‑specific hits and found none relevant.
• No proceeding numbers, panels, grounds, institution dates, FWD dispositions, settlements, or CAFC dockets are reported above, because none exist to report. Any such detail I might have written would have been fabricated.
• One cross‑reference to the earlier section: the observation that the granted claims were narrowed to vehicle‑autonomous execution during prosecution stands, and is consistent with there being no PTAB file wrapper. There is no contradiction between the two sections.

I'll verify the assignment records and look for correspondent/attorney-of-record details before writing up.

Inventors

Employer basis: every inventor's sole assignment is a whole‑right, title and interest transfer to International Business Machines Corporation, recited as "a corporation of New York having a place of business at Armonk, New York," recorded at reel 040704/0502 (rec. 2016‑12‑11). That assignment instrument is the standard IBM inventor‑to‑company form (the same form appears on other IBM dockets, e.g. reel 036804/0065 and reel 044799/0766), and its recitals state the assignee "desires to acquire" and the inventors "desire to grant." That is direct evidence the inventors were IBM personnel at filing; I could not determine from any retrieved record which IBM site/geography each inventor sat in — the reel/frame records do not carry subscriber addresses, and I did not obtain the underlying assignment PDFs.

Unusual-pattern check — all inventors departing within 12 months: Not determinable / no evidence. Assignment records capture conveyance of rights, not employment termination. None of the retrieved records (reel 040704/0502, reel 055912/0439, reel 057885/0644) contains an employment or departure date, and I found no news, SEC, or LinkedIn‑type evidence of bulk inventor departures. I am flagging this as unknown, not "not present," because the datapoint simply is not in this record type.

Original assignee

International Business Machines Corporation (Armonk, NY). Named as assignee on the issued patent and recorded at reel 040704/0502.

• Primary line of business: diversified enterprise IT — hardware, software, cloud, and services. At the relevant time (2016–2019) IBM was also a top‑5 annual US patent recipient.
• Did IBM ship a product embodying the claims? No evidence of one. The claims cover a vehicle‑resident risk‑analysis engine that reads vehicle sensors (body, proximity, daylight, rain, hand‑grip, gas/brake pedal), matches sensor data to risk patterns, contextualizes to a "personalized risk value," and has the vehicle autonomously execute a context‑modifying action (claim 1, granted form). IBM was never an automotive OEM and the specification's apparatus 10 is described as an in‑vehicle unit, but IBM did not commercialize it. This is a portfolio/research patent, not a product patent — relevant to the verdict below.
• Current status: operating (NYSE: IBM). Not acquired, dissolved, or in bankruptcy. IBM remains the assignor of record for the transfer below.

Assignment timeline

Three assignment documents are recorded against US 10,274,338, followed by two non‑assignment post‑issuance events.

• 2016‑11‑02 to 2016‑11‑22 (inventor execution dates) / recorded 2016‑12‑11 — Reel 040704/0502

  • Conveyance: Assignment (assignment of assignor's interest)
  • Assignor: Boss, Gregory J.; Cuervo, Ivonne R.; Hasbun Pacheco, Maria E.; Macias Garcia, Andrea; Navarro Piedra, Blanca R.; Zamora Duran, Edgar A.
  • Assignee: International Business Machines Corporation, Armonk, NY
  • Correspondent: Not retrieved. I could not obtain the assignment cover sheet for this reel/frame; the Assignment Center and legacy‑assignments PDF for 040704/0502 did not surface in my searches. Flagged as an open item — see signal 3 below.
  • Context: Initial inventor‑to‑corporate assignment — standard IBM employment‑docket capture; IBM Docket prefix on the sibling IBM instruments in this era is a YOR/ROC/IBA‑style identifier, consistent with an internally drafted instrument rather than a law‑firm recording.

• 2016‑11‑02 to 2021‑02‑11 (execution window) / recorded 2021‑02‑12 — Reel 055912/0439

  • Conveyance: Assignment — Corrective Assignment (corrects a typographical error in an inventor name previously recorded at reel 040704/0502)
  • Assignor: Boss, Gregory J.; Cuervo Fajardo, Ivonne Rocio; Hasbun Pacheco, Maria E.; Macias Garcia, Andrea; Navarro Piedra, Blanca R.; Zamora Duran, Edgar A.
  • Assignee: International Business Machines Corporation, Armonk, NY
  • Correspondent: Not retrieved.
  • Context: Administrative correction only — fixes "Ivonne R. Cuervo" → "Ivonne Rocio Cuervo Fajardo." No change in beneficial ownership. Note the odd execution window: the document was signed as late as 2021‑02‑11 on an assignment first recorded in 2016, i.e. a re‑execution to support the correction.

• Effective 2021‑09‑30 / recorded 2021‑10‑13 — Reel 057885/0644

  • Conveyance: Assignment (assignment of assignor's interest)
  • Assignor: International Business Machines Corporation
  • Assignee: KYNDRYL, INC., New York
  • Correspondent: Not retrieved.
  • Context: Corporate spin‑off / internal reorganization. This is the IBM→Kyndryl separation, not a sale to an acquirer. Critically, reel 057885/0644 is a bulk transfer: the identical reel/frame appears in INPADOC legal‑status records for unrelated IBM patents (e.g. US 10,277,521, whose record shows the same "ASSIGNOR:INTERNATIONAL BUSINESS MACHINES CORPORATION; REEL/FRAME:057885/0644" and effective date 20210930). A single reel/frame spanning an entire corporate portfolio is the signature of a spin‑off or re‑recordation, not of a targeted, patent‑by‑patent NPE acquisition.

Non‑assignment post‑issuance events (listed for completeness, not part of the chain of title):

• 2021‑04‑27 — Certificate of Correction (ex parte, USPTO‑issued).
• 2021‑05‑05 — Petition related to maintenance fees granted (event code FEPP).
• 2022‑10‑24 — 4th‑year maintenance fee paid, large entity (M1551). Fee current; term runs to the listed 2037‑05‑23 adjusted expiration.

Correspondent-of-record caveat (applies to all three links): the reel/frame data I could retrieve (Google Patents legal events, which mirror USPTO assignment data, plus INPADOC) does not expose the correspondent field. I did not fabricate one. For orientation only — and explicitly not attributed to this patent — IBM‑era assignments on other IBM dockets were routinely recorded by IBM's own IP law department (for example, reel 011994/0500 on an IBM Austin docket lists correspondent "CASIMER K. SALYS, 11400 Burnet Road, Austin TX" and is signed by IBM personnel), which is the expected pattern for an internally recorded corporate assignment. Verify against reel 040704/0502, 055912/0439, and 057885/0644 directly at assignmentcenter.uspto.gov (alternate front end: https://assignment.uspto.gov/patent/index.html).

Timeline diagram

timeline
    title Ownership of US 10274338
    2016 : Filed 11 Dec by IBM
         : Inventors assign to IBM
    2019 : Patent issues 30 Apr
    2021 : Corrective assignment recorded 12 Feb
         : IBM transfers to Kyndryl effective 30 Sep
         : Kyndryl assignment recorded 13 Oct
    2022 : 4th year maintenance fee paid

NPE / troll-pattern signals

Verdict

Insufficient data — no NPE pattern, no defensive‑aggregator terminus.

The complete chain is ordinary corporate: inventors → IBM (reel 040704/0502, recorded 2016‑12‑11), a name‑only corrective assignment (reel 055912/0439, recorded 2021‑02‑12), then IBM → Kyndryl, Inc. (reel 057885/0644, effective 2021‑09‑30, recorded 2021‑10‑13) as part of a bulk portfolio spin‑off — the same reel/frame appears on unrelated IBM patents such as US 10,277,521. That is a corporate separation, not a transfer‑to‑asserter, and none of the eight NPE/defensive signals is present. I land on Insufficient data only because the taxonomy offered has no bucket for "operating company owns it, ships no product, asserts nothing," and the one fact that would tighten the analysis — the correspondent of record on each reel/frame — could not be retrieved in this session; I decline to infer a name for it.

Verification: USPTO Assignment Center — search by patent number 10274338 (alternate front end: https://assignment.uspto.gov/patent/index.html). The three reel/frame entries to pull are 040704/0502, 055912/0439, and 057885/0644.

Caveat on sourcing, stated plainly: the reel/frame numbers, dates, and conveyances above come from the authoritative full text you supplied and from INPADOC/Google Patents legal‑status records that mirror USPTO assignment data. I could not open the Assignment Center record pages themselves, so the correspondent field is missing for every entry and no cover sheet was inspected. Nothing above should be read as a correspondent finding.

I'll verify the patent against USPTO-facing sources and then work through each cited reference.

Prior-Art Analysis — US 10,274,338 B2

0. Verification of the target patent number (literal, no auto-correction)

I ran the number 10274338 as given. Results returned the same document from multiple independent USPTO-derived sources, so this is not a look‑alike:

Corroborating hits: Google Patents (patents.google.com/patent/US10274338B2/en), Justia (patents.justia.com/patent/10274338), RPX Insight (insight.rpxcorp.com/patent/US10274338B2), and an appearance of US10274338B2 in the citation list of EP 1 328 421 B1. I excluded other 10,27x,xxx numbers that surfaced incidentally.

Transparency caveat: I could not open a raw USPTO PatentCenter / Global Dossier page for this number within my tool budget — my last calls were cut off by a step limit. The sources above are USPTO-derived mirrors, not the USPTO's own interface. Treat the bibliography below as accurate-to-the-record but not first-generation-verified.

1. Legal framework I am applying (and why it changes the answer)

Because the effective filing date is 2016-12-11, this patent is squarely AIA (post-16 March 2013). Two consequences matter for the question asked:

(a) Two different § 102 buckets exist. References published before 2016-12-11 are candidates under § 102(a)(1). References published on/after 2016-12-11 can only be art under § 102(a)(2) ("effectively filed" before) — and only if they are a U.S. patent, U.S. application publication, or WIPO publication. Several of the 21 cited references fall into this second bucket, which is a point the earlier sections of this analysis did not address.

(b) "Anticipation" is an all-elements test. A reference anticipates only if it discloses every element of the claim as arranged. Since claims 2–7 / 9–14 / 16–20 all depend from an independent claim, a reference that does not disclose all elements of claim 1 cannot anticipate any of the dependent claims either. So the honest answer to "which claims does reference X anticipate" is almost always "none, strictly."

Claim‑1 element checklist I used:

Elements G and I are the load-bearing, hardest-to-find limitations. Element D's "occupants" (plural, non-driver) is also distinctive.

2. Group A — Cited references published before 2016-12-11 (§ 102(a)(1) / § 103 candidates)

Dates below are the dates carried in the record (priority date first, then the publication/grant date as listed in the "Citations" table).

3. Group B — Cited references whose publication post-dates the 2016-12-11 filing (only § 102(a)(2) art)

This is the correction/refinement to the earlier sections: four of the twenty-one citations could not have been § 102(a)(1) art, because they published after the filing date. They are citable only because they were "effectively filed" earlier (pre-AIA "§ 102(e)" analog).

4. Non-patent literature cited on the face (6 items) — § 102(a)(1) printed publications

5. Verdict: the most relevant prior art, and what it does not reach

No cited reference anticipates claim 1 (and therefore none anticipates claims 2–7, 9–14, or 16–20, whether as independents or dependents). The single most likely examiner position for every one of the 21 citations is § 103, not § 102. That conclusion follows from the element checklist: no cited reference discloses the combination of (E) a risk‑pattern database, (F) a risk similarity value from pattern matching, (G) increasing that value using circumstance rates to reach a personalized risk value, and (I) autonomous vehicle-executed personalized action.

Ranked by relevance:

1. US 9,791,864 B2 (Ford) — best art on element I (automatic action via a vehicle control unit when a risk index crosses a threshold; action selection by greatest risk reduction). Weakest on A–C, E–G.
2. US 2015/0025917 A1 (Advanced Insurance / Stempora) — best art on element G's conceptual core (individualized, baseline-anchored cognitive/contextual risk scoring, plus vehicle-side responses and behavior modification). Weakest on E/F (no pattern DB, no similarity value) and on the autonomous-takeover framing.
3. US 2017/0057411 A1 (Intelligent Imaging Systems) — best art on the context-conditioned database + comparison-to-risk-criteria structure (C/F/H analogues). No rates, no similarity value, no takeover. § 102(a)(2) art.
4. US 8,892,451 B2 (Progressive) — best art on the sensor suite, data bus, and risk-level scoring plumbing (element D).
5. US 8,849,501 / US 8,508,353 (Lytx / Drivecam) — best art on automatic event scoring and calibration (supports the training-process dependent claims 6/13/19).
6. US 9,047,170 (Naboulsi) — secondary art on vehicle safety control acting on sensed operator state.
7. Krishnaswamy et al. (2005) and Rendon‑Velez et al. (2009) — NPL that establishes the situation-awareness/risk-prediction framing and supplies motivation to combine.

Conversely, references that cut against the patent's premise rather than toward it — worth noting because the specification expressly frames the invention as an improvement over them: US 8,508,351 (Bosch), US 2014/0240114 (Volkswagen), and the alert-centric parts of US 2015/0025917 all disclose warning the driver. The patent's own description distinguishes itself by not merely alerting a stressed driver (the spec: turn on soft music "instead of showing an alert which could stress him/her even more"). An examiner citing those references for the alert step would have to contend with that disclosure.

6. Contradictions / caveats to flag explicitly

• Contradiction with an assumption in the earlier-generated framing: the prior summary treated the cited-art set as uniformly "prior art." Literally read, four of the 21 citations (US 2017/0057411, US 2017/0076395, US 2017/0089710, US 9,791,864) published after the 2016-12-11 filing date and are therefore only § 102(a)(2) art. That is a substantive, not cosmetic, distinction.
• Date discrepancy reported literally, not corrected: US 2015/0025917's priority is listed as 2013-07-15 in the patent's own citation table but 2013-07-14 in one third-party mirror. I report both as found.
• "Potentially anticipates" is doing defensive work in the question. I have answered it strictly (all-elements test) rather than loosely. If the examiner's actual rejection grounds are what you need, the operative question is § 103 combinations, and the record shows no litigation, IPR, or PTAB proceeding for this number that would reveal a tested combination.
• I did not verify the references' specifications against the claim language from their full texts in every case — for the thinner references (Wilkes, Alert Enterprise, Wunelli, Digimarc, GM, Inrix, Allstate) I relied on the title/abstract-level descriptions carried in the record and in search snippets. Their element-by-element content should be confirmed against the primary documents before being relied on in a validity opinion.
• Forward citations are not prior art. The "Cited By" items (US 2021/0319238, US 12,198,448, EP 4,105,842, US 12,394,320, and the Kyndryl sibling US 11,299,174) post-date the patent and cannot be § 102 art; they are relevant only to later-art / family considerations.

I'll analyze obviousness under § 103, but first I need to ground the cited prior-art references with current search results. Let me pull details on the key references.

Note on a contradiction: The task header says "Current Date: April 26, 2026," while my operating instruction states today is 2026-09-29. I'll flag this rather than silently pick one; the earlier sections referenced "CAFC 2026" activity. I'll proceed and treat dates literally where they appear.

Good grounding. Let me pull the remaining key references.

Preliminary notes on constraints and conflicts

1. Date contradiction (flagged, not resolved): The task header states "Current Date: April 26, 2026," while my operating instruction says today is 2026-09-29. The previously generated "Patent summary" section also referenced "CAFC 2026" activity while elsewhere concluding no litigation was found. I do not reconcile these; I note them so you can treat any date-stamped conclusion accordingly.
2. Search budget exhausted mid-task. I completed grounding searches for US 2017/0057411 (Intelligent Imaging Systems), US 8,849,501 / US 8,508,353 (Lytx/DriveCam), US 9,047,170 (Naboulsi), US 9,791,864 (Ford), and US 2017/0089710 (Allstate). Searches for the remaining cited references were cut off by the step limit. For those, I rely on (a) the titles/assignees/dates as recited in the '338 patent's own "Citations (21)" list (which is authoritative for what was cited, if not for what it discloses), and (b) my prior knowledge. Anything I say about the unretrieved references is lower-confidence and should be independently verified.
3. Statutory framework: The '338 patent was filed 2016-12-11, so the AIA versions of §§ 102/103 govern. All references below qualify as prior art under § 102(a)(1) (patents/printed publications before the filing date) or § 102(a)(2) (U.S. application publications "effectively filed" before the filing date — relevant for US 2017/0057411, US 2017/0076395, US 2017/0089710, and US 9,791,864, each of which published after 2016-12-11 but had earlier effective filing dates).

I. The claim as the obviousness target

Independent claims 1/8/15 share nine steps. The legally salient limitations are:

PHOSITA (2016): a person or team with a bachelor's in EE/CS/ME (or equivalent) and ~2–3 years in automotive telematics, sensor fusion, or driver-assistance systems, familiar with on-board diagnostics, ADAS/ECU control, and machine-learning scoring.

Note: the granted claims (with "autonomously executing by the vehicle … take over") are narrower than the published application. The obviousness case must therefore supply both (d) contextual personalization and (f) autonomous take-over.

II. What the retrieved references actually teach

US 2017/0057411 A1 — Intelligent Imaging Systems, "Contextual driver behavior monitoring" (priority 2015-08-27; pub. 2017-03-02; granted as US 9,849,887)

• Maintains a database of "high risk locations," each "associated with driver behaviors that are considered high risk," then "compar[es] the driver behavior data to the driver behaviors considered high risk … to determine a value or values representing driver risk, and flagging a driver whose driver risk value or values exceed predetermined risk criteria." → elements (c)+(e).
• Its stated rationale is contextualization: it "compares it to data from other drivers and onboard sensor data to accurately identify high risk driving behavior"; "Random deviations of uncontextualized vehicle events have a tenuous association with driver safety identification."
• Critically, it identifies the very gap the '338 patent purports to fill: "the state of the art contains little consideration of instantaneous driver coaching." Its disclosed output, however, is a visual warning, not autonomous take-over. (freepatentsonline.com/y2017/0057411.html)

US 8,849,501 B2 — Lytx, "Driver risk assessment system and method employing selectively automatic event scoring" (granted 2014-09-30)

• Onboard detectors "record data related to detected driving events, vehicle condition and/or tasking, roadway environmental conditions," and the system "creates a driver score based not only upon the frequency and severity of driving events, but also adjusted (if appropriate) for different external factors"; the score is "normalized by consideration of environmental factors." → element (d)'s contextual adjustment of a risk score.
• The onboard system will "score a detected driving event, compare the local score to historical values previously stored within the onboard system, and upload selective data … if the system concludes that a serious driving event has occurred." → (c)/(e) plus the feedback loop of claims 2/3.
• Events include "reckless driving." (patents.justia.com/assignee/lytx-inc)

US 8,508,353 B2 — DriveCam/Lytx, "…having calibrating automatic event scoring" (granted 2013-08-13)

• Calibration of event scoring from accumulated data → supports the training process of claims 6/13/19. The related Lytx "engine" literature describes "real-time decision algorithms that continuously monitor the sensor stream to determine the likelihood of risky driving behaviors." → claim 5.

US 9,047,170 B2 — Naboulsi, "Safety control system for vehicles" (granted 2015-06-02)

• A "controller … to selectively suppress at least one of said input and said output in response to a sensed parameter … being outside of a threshold." → element (f) (vehicle autonomously alters/inhibits vehicle functions upon threshold).
• Explicitly senses driver stress physiologically: steering-wheel sensors detect "a predetermined gripping force … pulse rate, temperature, blood pressure, and/or skin conductivity"; "Such physiological condition may indicate a stress condition of the driver and, when sensed, disable … the telephone so as not to aggravate the stressed condition." → claim 4's "mood of the driver." (patents.google.com/patent/US9047170B2)

US 9,791,864 B2 — Ford, "Systems and methods for driving risk index estimation" (granted 2017-10-17; priority 2016-03-10)

• "calculate a first risk estimation … in response to the first risk estimation satisfying a risk threshold: determine a risk reduction action; and instructing a vehicle control unit to automatically implement the risk reduction action." → elements (e)+(f) squarely.
• The vehicle "automatically (e.g., without driver intervention) takes an action to reduce the estimated risk index," and "implements the action that reduces the risk index the most." → the take-over limitation and the "select best action" idea of the spec. (patents.google.com/patent/US9791864B2)

US 2017/0089710 A1 — Allstate, "Three-Dimensional Risk Maps" (filed 2015-09-24; pub. 2017-03-30; granted as US 11,307,042)

• Computes a risk score, then "calculate a modified risk value based on environmental information" — e.g., "adjust a risk value due to a new condition (e.g. snow on the road)" using a "multivariable equation."
• "determine if the risk value … is over a threshold"; if so, "update the risk map"; and it uses "pattern matching" (step 725).
• Receives driver-side context ("user's age, gender, marital status … eyesight … physical disability"), weather, geographic location, and vehicle info.
• It can "influence autonomous and semi-autonomous vehicles … to slow the vehicle down or to help the vehicle avoid an accident." → (d)+(e)+(f). (freepatentsonline.com/y2017/0089710.html)

Unretrieved (lower confidence — verified only as cited by the '338 list): US 6,335,689 (Fuji Jukogyo, driver arousal-level estimation), US 6,925,425 (Motorola, operator performance assessment/improvement), US 8,508,351 (Bosch, warning of increased accident risk), US 2017/0076395 (Inrix, driving-behavior/risk rating), US 2005/0174217 (Basir, recording/reporting driving characteristics), US 2010/0077028 (Wilkes, smartphone safety platform), US 2014/0240114 (VW, driver-assistance alerts), US 2014/0375810 (Digimarc, vehicular safety), US 2015/0025917 (Advanced Insurance, underwriting risk using "cognitive information"), US 2015/0325121 (GM, decision support), WO2014016620 (Wunelli, driving-behaviour monitoring), WO2011063269 (Alert Enterprise, risk visualization/remediation). Non-patent: Remboski et al., "Driver Performance Improvement Through the Driver Advocate" (SAE 2000); Rendon-Velez et al. survey (2009); Krishnaswamy et al. (2005).

III. Obviousness grounds

Ground 1 (strongest): Intelligent Imaging '411 + Lytx '501 + Ford '864 (or Naboulsi '170)

Motivation: All three are in vehicle-safety telematics; each expressly addresses the same problem (detecting risky driving before a crash). Intelligent Imaging '411 itself supplies the why — it laments the absence of "instantaneous driver coaching." Lytx '501 supplies contextual normalization. Ford '864 supplies the automation layer, and Ford teaches that a risk index crossing a threshold should trigger ECU action. Combining a contextual/scoring engine with an autonomous control layer is the predictable use of a known technique to improve a similar device (KSR rationales (A), (C), (D), (F)).

Ground 2: Allstate '710 + Ford '864 + Lytx '501

Allstate '710 may be the closest single-art reference: it computes a risk score, modifies that score based on context (snow), compares to a threshold, uses pattern matching, and can "influence autonomous and semi-autonomous vehicles … to slow the vehicle down." Ford '864 closes the take-over with explicit ECU instruction. Lytx '501 supplies the driver/occupant identity and "environmental factor" normalization. Under KSR, "if a technique has been used to improve one device, and a person of ordinary skill … would recognize that it would improve similar devices in the same way, using the technique is obvious."

Ground 3 (physiological/mood-context channel): Naboulsi '170 + Fuji '689 + Motorola '425 + Ford '864

For the dependent-claim features (driver mood, profile, training), Naboulsi '170 (stress via pulse/grip), Fuji '689 (arousal level), and Motorola '425 (operator performance assessment) collectively teach deriving driver state from sensors and adjusting system behavior. Ford '864 supplies threshold-triggered automated action. This ground is most useful against claims 4/11/18 and 6/13/19 rather than claim 1's core.

Ground 4 (non-patent literature): Remboski "Driver Advocate" (SAE 2000) + Rendon-Velez survey (2009) + Krishnaswamy (2005) + Ford '864

The Remboski "Driver Advocate" initiative is cited on the '338 face and, as I understand it (lower confidence), describes an adaptive in-vehicle system that monitors driver state and context and coaches/intervenes to improve performance. The Rendon-Velez survey expressly frames "situation assessment and risk prediction in advanced driver assistance systems." These establish that context-aware, personalized driver-risk scoring with automated intervention was a recognized research program well before 2016. Combined with Ford '864's automation, they support the "obvious to try" rationale over a finite set of identified approaches.

IV. The dependent claims

• Claims 2/9/16 and 3/10/17 (loop-back on non-exceedance / no match): routine programming; Lytx '501's continuous scoring loop.
• Claims 4/11/18 (driver ID/profile/mood/time/weather/GPS/route): Allstate '710 (age/gender/disability, weather, GPS, route), Naboulsi '170 (stress physiology), Motorola '425 / Fuji '689.
• Claims 5/12 (continuous sensor collection): Lytx continuous sensor-stream monitoring.
• Claims 6/13/19 (training: default context/rates/patterns, then adjust rates): Lytx '353 ("calibrating automatic event scoring") and the Lytx machine-learning literature; Wunelli WO2014016620 (lower confidence).
• Claims 7/14/20 (context modifier changing environmental variables): the '338 spec's own examples ("open windows," "radio volume") map onto known infotainment/HVAC actuators; Naboulsi '170's suppression is a species of context modification.

V. Where a patentee could push back (honest assessment)

1. The specific "increase the risk similarity value" arithmetic. Claim 1 requires that contextualization increase a pre-computed similarity value (rather than compute risk from scratch). Lytx '501's "normalized by consideration of environmental factors" and Allstate '710's "modified risk value" are close, but whether they increase a match-similarity score derived from pattern matching (versus adjusting inputs to a fresh score) is a genuine factual question I cannot resolve from the retrieved text alone.
2. Take-over vs. warning. Several references (Bosch '351, Intelligent Imaging '411, Lytx) are warning-centric; a patentee will stress that the granted claims were narrowed toward autonomous vehicle take-over, implying the examiner credited that limitation. The Ford '864 and Naboulsi '170 disclosures are the strongest rebuttal, but they must be shown to be combinable in the specific integrated architecture.
3. Secondary considerations. If Kyndryl can show unexpected results or industry praise for the personalization mechanism, that weighs against obviousness. I found no evidence of either.
4. Prosecution history. I do not have the file wrapper for US 15/375,136, so I cannot identify what the examiner actually relied on or what applicant argued. Notably, the examiner cited all 21 references yet allowed the case — suggesting the combination was found non-obvious, which is precisely what these grounds must overcome.

VI. Bottom line

On the retrieved record, claims 1/8/15 are, in my assessment, at substantial risk of invalidation under § 103 via the combination of a contextual driver-risk-scoring reference (Lytx '501 or, especially, Allstate '710) with an autonomous-intervention reference (Ford '864 or Naboulsi '170), with Intelligent Imaging '411 supplying the explicit motivation to move from generalized risk profiling to instantaneous, contextual intervention. The dependent claims add little, being either routine implementation details or features distributed across the cited physiological/context references.

Two caveats bound this conclusion: (1) I could not verify the substantive disclosures of roughly a dozen cited references, several of which (Bosch '351, Motorola '425, Fuji '689, Wunelli) may independently strengthen or weaken specific grounds; and (2) my conclusion is based on the references' apparent disclosures, not on an evidentiary record, and a formal invalidity contention would require full-text review of each reference plus the '338 prosecution history.