For practicing clinicians

A second pair of eyes that has read every page.

ScoutMyCase is a source-backed clinical reasoning workspace that ingests a patient's longitudinal records, builds a canonical fact graph, and ranks candidate trials and treatments against current eligibility — with provenance for every claim.

Try a sample case Read the methods
Source-backed reasoning Open-source methodsEncrypted & access-controlled
Eligibility check · NCT06556563
T. Foster, 51 — glioblastoma, IDH-wildtype, MTAP-deleted
94%
✓
Glioblastoma, IDH-wildtype
IU Health pathology, Feb 19 2026 · path #PA-26-0447
✓
MTAP / CDKN2A homozygous deletion
Caris MI Cancer Seek, Mar 1 2026
✓
Newly diagnosed, post-resection
Craniotomy + resection, Feb 17 2026
✓
KPS ≥ 70
Oncology note, Feb 24 2026 · KPS 80
✓
Adequate organ function
Pre-op CBC/CMP within range
!
Enroll before chemoradiation — closes ~1 week
Radiation planning underway
Every claim shows its source.Open packet
Vs. searching ClinicalTrials.gov yourself

We're not matching a similar case. We're checking every detail of this case.

A keyword search returns trials that look like the diagnosis. Real eligibility lives 14 layers down — in a path report, a discharge summary, a 2019 sequencing panel, an ECOG score from a note three months back. Doing it by hand for one patient takes hours. We do it for every open trial, and we show our work.

01 · Manual
Manual review
A clinician scans 2–3 trials by name, glances at top-line eligibility, takes 20–40 minutes per patient.
02 · The default
Keyword matchers
Surface trials by diagnosis match. Eligibility stays a footnote — most "matches" fail at screening.
03 · What we do
ScoutMyCase
Every criterion of every open trial, checked against the longitudinal record. Each pass or fail cites the source document and date.
One patient, one trial, 47 criteria

What "checking every detail" actually means.

Trial protocol asks
  • 01Histologically confirmed glioblastoma, IDH-wildtype
  • 02MTAP and/or CDKN2A homozygous deletion
  • 03Newly diagnosed, after maximal safe resection
  • 04No prior anti-cancer therapy for glioblastoma
  • 05KPS ≥ 70 at screening
  • 06Adequate hematologic, hepatic, renal function within 14 days
  • 07Enrolled before chemoradiation begins
  • 08… 40 more criteria
Where we found the answer
  • Surgical pathologyFeb 19 2026 · IU Health · GBM, IDH-wildtype
  • Caris MI Cancer SeekMar 1 2026 · MTAP/CDKN2A deletion
  • Operative noteFeb 17 2026 · craniotomy + resection
  • Cross-record reviewIU Health + Caris · no prior therapy
  • Oncology noteFeb 24 2026 · documented KPS 80
  • Pre-op lab panelFeb 2026 · CBC + CMP within window
  • Radiation planningPending start · window ~1 week
  • 46 documents indexedacross IU Health + Caris
How a case moves through it

From scattered records to a traceable next action.

1
Ingest
Records from any portal, fax, or upload. PDF, DICOM, HL7, FHIR.
2
Normalize
Extract facts into a canonical graph with provenance per claim.
3
Find gaps
Flag missing facts that change ranking — and frame them as questions.
4
Match
Score every open trial against the full record. Surface blockers, not just hits.
5
Hand off
A one-page packet with sources, eligibility, and the questions to ask.
What you can inspect

Audit any answer to its source.

Nothing is hidden behind "the model said so." Each claim points to a document, a page, and a sentence. You can disagree, and we'll show you the reasoning trace.

Open the methods
Canonical graph
Every fact extracted from records, with type, value, source document, page, and confidence.
FHIR-ish import & export
Import bundles from major EHRs. Export the case as FHIR Resources for downstream use.
Eligibility traces
For each criterion: the rule, the matching fact, the source, and the decision logic.
Open methods
App code, runtime, schemas, prompts, and tests are public. Self-host if you prefer.
Negative evidence
When something doesn't fit, you see why — not just what fits.
Living watchlist
Re-rank when new evidence publishes or trials open. You're notified, with diff.
On scope

A research aide. Not a recommender. Not a doctor.

We surface options, flag mismatches, and show our work. Decisions stay where they belong — with you and the patient. We will never produce a "the answer is X" without provenance, and we will tell the patient when we're uncertain or out of our depth.

Clinical advisory

Built with practicing clinicians.

Specialists across neuro-, thoracic, and hematologic oncology, primary care, palliative medicine, and clinical informatics shape what we recommend, how we explain uncertainty, and what we refuse to say.

Dr. M. Konecny, MD
Neuro-oncology
Northwestern
Dr. R. Adeyemi, MD, MPH
Thoracic oncology
MSK
Dr. S. Vance, MD
Hematology
BIDMC
Dr. P. Iyer, MD
Primary care, internal medicine
Mass General
Dr. J. Larkin, MD
Palliative medicine
UCSF
Dr. A. Roth, MD, PhD
Clinical informatics
Stanford
Common questions

What clinicians ask first.

A keyword search returns trials whose title or condition matches the diagnosis. We score every open trial against every criterion — using the longitudinal record, not the visit summary — and surface the blockers with sources. Most "matches" from a keyword search fail at screening; ours tell you why before you refer.

Open a sample case.
See what we'd hand you.

Tim Foster is a real shape of patient — newly diagnosed glioblastoma, IDH-wildtype, MTAP-deleted, with records from surgery and molecular testing. Walk through the case the way one of yours might.

Try a sample case Talk to the team