More Complete Patients. More Confident Research
Researchers build cohorts from structured EHR data. But the details that define a patient — smoking history, disease stage, performance status, biomarker results — live in clinical notes. Facts from Clinical Notes makes them accessible.
The cohort you’re missing
Structured EHR data is the foundation of real-world research. Diagnoses, procedures, medications, labs, and demographics are the backbone of defining patient cohorts. But the facts that differentiate one patient from another — their disease stage, severity, performance status, biomarker results, response to therapy, functional decline, and social determinants — can be recorded as narrative text in notes.
From narrative text to research-ready facts
Facts from Clinical Notes is a validated set of clinical facts extracted from unstructured notes recorded at the point of care that don’t make it into a structured field. Using deep learning models with clinicians in the loop, those notes are transformed into high-confidence, validated facts that complement the structured EHR data researchers already rely on.
Coverage spans oncology, pulmonology, cardiology, biomarkers, liver disease, alcohol use, and more. Note volume, patient coverage, and fact types expand over time, with new therapeutic areas prioritized by customer feedback — so the data keeps pace with the questions you need to answer.
6x More Patients
Clinical notes delivered 6x more patients in a lung cancer cohort study.
3x More Patients
Smoking status coverage increases from 22% to 67% when notes-derived facts are included.
Details Structured Data Can't Capture
0 patients to 156,000 patients with oncology performance status (ECOG/Karnofsky) — data that doesn’t exist in structured fields.
5x More Facts
Pulmonary function tests extracted from notes increased fact count by 5x and doubled patient coverage.
Not just better data, but better access to it
Customers receive validated facts, not a corpus to process or a platform for building their own extraction. No internal NLP capability is required.
Facts from clinical notes appear in TriNetX LIVE™ next to the structured EHR data for the same de-identified patients. Cohort building and analysis work the way they already do. No new tool, no new login, no separate workspace.
The same validated facts are available in downloadable datasets for offline analysis, regulatory submissions, and custom analytical workflows.
Data is sourced directly from participating healthcare provider sites and kept within those institutions. The breadth and institutional diversity of the network is something a centralized repository can’t replicate.
Note volume, patient coverage, and fact types expand over time, with new therapeutic areas prioritized by customer feedback.
Build the evidence your structured data can’t
Your cohorts may be smaller than they should be — not because the patients aren’t there, but because the facts that define them were documented in notes, not coded into structured fields. For clinical operations teams, that means feasibility counts built on structured data alone risk undercounting the patients who are actually recruitable. For research teams, it means cohorts defined on coding proxies rather than clinical reality, and evidence that doesn’t carry the precision regulatory and payer audiences expect. Facts from Clinical Notes makes those criteria accessible alongside structured data, so every decision — from protocol design to label expansion — starts from a population that actually exists.
- Which sites treat the patients my protocol actually requires, based on what clinicians documented?
- What biomarkers are present in my target population, and how does that change my eligible cohort?
- How are patients being managed in practice, beyond what’s coded in claims?
- What does disease progression look like in the real world, from diagnosis through therapy sequencing?
- How does functional status vary across patient subgroups, and what predicts treatment switching or discontinuation?
The scale and depth of clinical notes data without the NLP project
Single-institution patient pools are rarely large enough to power statistically meaningful findings, and structured EHR data leaves the phenotypic detail that defines those patients buried in free text. Getting value from clinical notes has historically meant months of IRB negotiation, data use agreements, and NLP pipeline development before a single analysis could begin. Facts from Clinical Notes removes that burden. Extraction and validation are already done, delivered query-ready in TriNetX LIVE™ across a network no single institution can replicate on its own — so research teams spend time on the work, not on the data preparation that precedes it. To understand how clinical notes data is extracted, validated, and made accessible — and what that means for the questions you can now answer — read the eBook.
- Powering studies that require sample sizes and clinical specificity beyond what a single institution can supply
- Building grant applications with population-level burden-of-illness evidence in days, not months
- Cohort identification and patient characterization using criteria that structured data alone can’t express
- Longitudinal analysis across disease progression, treatment sequencing, and real-world outcomes
- Feasibility assessments that reflect the full documented patient population, not just what was coded
- Validating single-institution findings against a broader real-world population
A growing corpus of facts
Facts from clinical notes are available now for a growing set of therapeutic areas. New indications are added each quarter.
Available now:
- Smoking: Status, Type, Pack Year, Duration, Frequency, Quantity
- Pulmonology: PFT, FEV1, FVC, FeNO
- Oncology: Performance Status, Histological Type, Grade, Stage (clinical & path)
- Biomarker Lung: PD-L1, EGFR, ALK
- Biomarker Colon: MSI, KRAS/NRAS
- Biomarker Breast: ER, PR, HER2, Ki-67, BRCA ½
- Biomarker Prostate: Gleason Score, PSA
- Cardiology: QT Interval, Ejection Fraction, NYHA, ACC, ASCVD
More coming soon across oncology, SDoH, autoimmune diseases, and more.
See what’s possible for your research
"*" indicates required fields
