Fastest
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →https://www.patsnap.com/resources/blog/rd-blog/admet-prediction-models-patent-landscape/ · Patsnap · data cut-off 2026-07-31 · downloaded from the live page
Patent Landscape · Computational Drug DiscoveryADMET Prediction Models: Patents Mapping Who Files and Where the Gaps Sit
- Concentrated at the top. The top 5 assignees account for 57.9% of all 19 records in scope, and the top 10 reach 84.2% — a short list of repeat filers surrounded by a long tail of single-filing entrants.
- Computational chemistry dominates the claim language. G16C appears in 78.9% of records, with G06N (AI models) at 47.4% and G16B (bioinformatics) at 31.6% — most filings pair a chemistry framing with a machine-learning method claim.
- Filing is still ramping, not plateauing. Volume rose from zero in 2017 to a peak of 6 in 2025, with 5 already published in the partial 2026 year — publication lag means the true 2026 count will land higher.
19
Published Records
58%
Top-5 Share of All Records
US
Leading Jurisdiction
28
Active Filers Ranked
Top-5 share is the combined record count of the five largest assignees divided by all 19 records in scope (CR5), not by the ranked leaders only.
Published byPatsnap Research··7 min readSourced from Patsnap Eureka
Overview
What this landscape covers
ADMET prediction models sit at the intersection of computational chemistry and machine learning: patents in this set claim methods for predicting absorption, distribution, metabolism, excretion or toxicity of candidate compounds, often anchored to applicability domain, uncertainty estimation or temporal validation split concepts that distinguish a defensible model claim from a generic QSAR method. The search scope pulls 19 published records between 2015 and the 2026 cut-off, filed by 28 ranked assignees spanning corporate, university and individual inventors.
The dataset is small enough that individual filings move the concentration figures meaningfully, so read the ranking as a signal of who is active rather than a mature competitive map. Receiving-office spread across the United States, China, India, WIPO, Europe and Taiwan suggests the field is being claimed in parallel jurisdictions rather than consolidated behind a single filing strategy.
Filing activity and technology composition, 2015–2026
Source: Patsnap Eureka. Assignee ranking and totals. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
10,000 free credits to start
The Data
Filing trend and technology composition
Two views of the same 19 records: filing activity by year, and how those records distribute across IPC subclasses. Because a single record can carry several IPC codes, the subclass shares below sum to more than 100% of the record total.
Filing trend, 2017–2026
Filings moved from zero in 2017 to a peak of 6 in 2025; 2026 already shows 5 published records but the year is only partial and publication lag of roughly 18 months means the final count will sit higher once late filings clear examination and publication.
IPC subclass composition
G16C (computational chemistry) covers 78.9% of the 19 records, far ahead of G06N (AI computing methods) at 47.4% and G16B (bioinformatics) at 31.6%. Smaller shares in C08L, G16H, C12Q and G01N show the technology reaching into polymer science, healthcare informatics and lab measurement, but each at only one or two records.
Shares are the percentage of the 19 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Source: Patsnap Eureka. Filing trend and technology composition. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.
Go deeper on ADMET Prediction Models with Eureka
This page is one run against one query. Ask Eureka your own question about admet prediction models and every answer comes back with the patent numbers behind it.
Try EurekaKey Patents
Most-cited records in scope
Representative Filing
US20250157590A1 — Systems and methods for chemical toxicity prediction
NATIONAL TAIWAN UNIVERSITY
The present disclosure relates to systems and methods for chemical toxicity prediction. The methods comprise receiving test data from an analysis instrument, selecting a candidate chemical according to the test data, determining a hazard translated level and a hazard evaluation level of the candidate chemical according to a molecular fingerprint, and predicting toxicity using a quantitative structure-activity relationship (QSAR) model based on the hazard translated level and the hazard evaluation level.Filed by National Taiwan University, published 2025-05-15.
Highest-citation records
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | CN118506882A | 一种用于预测非化疗药物诱发粒细胞缺乏症的毒性预测模型及其应用 | 3 |
| 2 | EP4456077A1 | Machine learning for property prediction and recipe recommendation of polyolefin-based polymer blends | 2 |
| 3 | US20250384966A1 | Systems and Methods for Automated Compound Synthesis | 1 |
| 4 | US20250157590A1 | Systems and methods for chemical toxicity prediction | 1 |
| 5 | WO2024223845A1 | Machine learning for property prediction and recipe recommendation of polyolefin-based polymer blends | 1 |
Citation counts inside this corpus are low in absolute terms and skew toward earlier-published records — read them as a signal of relative influence within this 19-record set, not as a measure of current commercial importance.
Patent titles are shown in the language they were filed in, not translated, so that each record stays verifiable against the original filing — a translated title will not match in Eureka or in any national register. Publication numbers are shown where the record carries one (5 of 5 rows); clicking a row searches Eureka by that number.
Source: Patsnap Eureka. Citation counts and representative records. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Run it yourself
Put your own technology through the same analysis
Where to run it
For builders
MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →Insights
What the numbers mean for a filing decision
Three read-outs from the concentration, composition and geography data, aimed at where a new filing would land relative to existing claims.
Concentration
57.9% / 84.2%
top 5 / top 10 share of 19 records
A short leader group, a long tail behind it
The top 5 assignees hold 57.9% of all 19 records in scope and the top 10 reach 84.2%, leaving the remaining 18 ranked assignees splitting a handful of filings between them. That pattern is typical of an emerging niche: a few organisations built early portfolios while most entrants have filed once.
Read the ranking as activity, not market share.
Claim overlap
78.9% G16C
share of records carrying computational chemistry class
Method claims cluster around chemistry-plus-ML framing
Nearly four in five records carry the G16C computational chemistry class, and almost half also carry G06N for AI computing methods. New filings that claim only a generic machine-learning pipeline without a chemistry-specific structure or descriptor risk sitting on the same crowded ground as the majority of existing claims.
Bioinformatics (G16B) is present in under a third of records — a comparatively lighter-claimed adjacency.
Jurisdiction spread
US 5 · CN 4 · IN 4
records by receiving office
No single office dominates filing strategy
Receiving offices split across the United States, China and India at similar volumes, with WIPO, EPO and Taiwan filings behind them. That spread suggests applicants are pursuing parallel regional protection rather than a single home-market strategy, which matters for freedom-to-operate checks that only clear one jurisdiction.
A single-country clearance search would miss most of this set.
Eureka AI Agent
Looking for what nobody has claimed yet?
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to admet prediction models, with the prior art for and against each one.
Source: Patsnap Eureka. Co-assignee relationships and derived observations. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Players
Who is filing, and where the gaps sit
The ranked assignee list spans corporate labs, universities and individual inventors, with recent-year activity concentrated in a small number of names.
Recent momentum
2 in latest year
SB Tech Inc, most recent year
One assignee is filing now while earlier leaders go quiet
SB Tech Inc shows 2 filings in the latest year, while several assignees that were active earlier — including DeepCure Inc and a Chinese university co-filer — show 0 filings in the latest year and a -100% year-on-year change. That handoff pattern is worth watching before assuming any single leader's position is stable.
Momentum in a 19-record set can shift on one or two filings.
Co-filing
3 pairs
co-assignee pairs identified
Co-assignments are rare but concentrated around university-agency links
Only 3 co-assignee pairs appear in the dataset, the strongest being a national environmental research body paired with a national university, filed twice together. That pairing points to a public-research collaboration model rather than corporate joint filing, which is otherwise absent from this set.
Co-filing here signals institutional partnership, not licensing activity.
Entity mix
28 ranked assignees
across the whole ranking
Individual inventors sit alongside corporate and university filers
The ranking includes named individual inventors alongside corporate entities such as Borealis and DeepCure and university filers such as National Taiwan University. That mix is unusual for a field this technical and suggests filing costs are low enough, or the claim scope narrow enough, that solo inventors are entering alongside institutional players.
Diligence on this field should not assume corporate-only competitors.
Recent-year filing momentum by assignee
| Assignee | Recent year | YoY |
|---|---|---|
| SB TECH INC | 2 | — |
| National Environmental Research Academy | 0 | — |
| National Taiwan University | 0 | — |
| Borealis AG (Austria) | 0 | — |
| DEEPCURE INC | 0 | -100% |
| Zhejiang Gongshang University | 0 | -100% |
| Jieyang People's Hospital | 0 | — |
| Douyin Vision Co., Ltd. | 0 | — |
Source: Patsnap Eureka. Assignee-level momentum. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
What's Next
Where to take this next
The concentration and composition figures point to specific next steps depending on whether the goal is freedom-to-operate, portfolio strategy or new filing.
Check freedom-to-operate across all three lead jurisdictions
With filings split across the United States, China and India at similar volumes, a clearance check limited to one office will miss active claims in this set.
Run a jurisdiction-specific searchDraft around the chemistry-plus-ML claim pattern
Since 78.9% of records carry the G16C computational chemistry class often paired with G06N, claims that combine a generic ML pipeline with a specific chemical descriptor structure sit on less-contested ground.
Explore claim drafting in EurekaTrack the SB Tech Inc filing pace against the quieter incumbents
Recent-year momentum has shifted toward SB Tech Inc while earlier active filers show no latest-year activity, a pattern worth monitoring before assuming portfolio stability.
Set up assignee monitoringSource: Patsnap Eureka. Forward-looking reading of the same dataset. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
FAQ
Common questions on ADMET prediction model patents
The scoped dataset contains 19 published records between 2015 and the 2026 cut-off, filed by 28 ranked assignees. This is a narrow field by patent-count standards, which means individual filings can shift concentration figures noticeably from year to year. Treat the count as the current state of a still-emerging niche rather than a mature, saturated technology area.
Filing is concentrated at the top: the five most active assignees account for 57.9% of all 19 records in scope, and the top ten reach 84.2%. Beyond that group sits a long tail of assignees with a single filing each, including universities, a national research agency and individual inventors. Recent-year data shows some momentum shifting toward newer entrants while earlier leaders show no latest-year filings.
Nearly four in five of the 19 records (78.9%) carry the G16C computational chemistry classification, and just under half (47.4%) also carry G06N for AI-based computing methods. Smaller but present adjacencies include bioinformatics (G16B, 31.6%), general digital data processing (G06F, 21.1%), and narrower slices in polymer compositions, healthcare informatics, enzyme/DNA testing and material analysis. The pattern shows most claims frame prediction as a chemistry-specific method rather than a generic machine-learning application.
US20250157590A1, filed by National Taiwan University and published 2025-05-15, claims a system and method for chemical toxicity prediction that receives test data from an analysis instrument, selects a candidate chemical, determines a hazard translated level and hazard evaluation level from a molecular fingerprint, and predicts toxicity using a QSAR model built on those two levels. The claim structure ties the prediction step specifically to the hazard-level determination sequence, not to QSAR modelling generally. Anyone building a toxicity predictor should check whether their pipeline reproduces that specific translated-level-plus-evaluation-level sequence before assuming clearance.
Sub-areas tied to model reliability rather than raw prediction — applicability domain boundary estimation, temporal validation split methodology, assay data heterogeneity correction, and model maintenance or drift monitoring — appear far less often in the claim language than core computational-chemistry prediction methods. Because these concepts are named in the search scope but not heavily represented in the 19 records, they represent claim space that is discussed in the literature but not yet densely filed. A first claim there would need to tie a specific validation or maintenance mechanism to a defined model architecture rather than claiming the concept generically.
Answers are grounded in the same dataset. Derived from a Patsnap search on ADMET Prediction Models covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Research ADMET Prediction Models in depth with Eureka
Go past this page: query the whole admet prediction models corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company's registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.