| name | chart-review-lung-cancer-adherence |
| description | Lung cancer (NSCLC) molecular-testing guideline-concordance review. Activate when assessing whether an NSCLC patient's molecular workup adhered to NCCN — genomic testing performed, comprehensive genomic profiling vs single-gene, PD-L1 IHC, testing-before-therapy sequencing, essential-biomarker completeness, and targeted-therapy alignment for actionable mutations — answering the MT0–MT12 chart-review questions per patient from pathology, molecular/genomics, radiology, and oncology notes plus structured EHR data.
|
Lung cancer molecular-testing — chart review + NCCN concordance
For each in-scope NSCLC patient you answer the FACTUAL questions in
references/questions/ (the MT0–MT12 checklist). The NCCN molecular-testing
concordance verdicts (C1–C6) are then computed deterministically by the
rule-engine from those facts (references/rules/concordance.yaml) — you do not
judge concordance directly.
Procedure
- Eligibility + context (
eligibility.yaml, T0): confirm a lung cancer
diagnosis, the histologic subtype (SCLC vs the NSCLC subtypes), the stage, and
smoking history. Answer first — histology + stage gate the concordance rules
(SCLC is generally not required to have molecular testing).
- Molecular workup (
testing.yaml, T1–T3): was genomic testing performed;
CGP vs single-gene/panel; tissue vs liquid; was testing ordered / resulted
before first-line therapy; was PD-L1 IHC performed + its result; the actionable
findings (gene + variant) and whether any has an FDA-approved targeted therapy.
- Treatment alignment (
treatment.yaml, T4): targeted therapy for actionable
mutations, time from diagnosis to first-line therapy (days), genomic turnaround
(days), and minimum-biomarker completeness when CGP was not done.
- Concordance is automatic. The rule-engine evaluates the seven NCCN rules
over your answers — each resolves to Concordant / Non-concordant / Excluded
(key documentation absent), and a non-concordant verdict gets a lung-specific
attribution category (
references/attribution.yaml). You do NOT answer these;
if a verdict looks wrong, correct the underlying abstracted fact.
- Cite evidence for every answer: a verbatim note quote (pathology /
molecular / radiology / oncology), or a structured row (labs/orders) when
that is the source. Answer with the exact enum token; put detail (vendor, gene
names, dates) in the rationale. Do not infer from parametric knowledge.
Concordance rules (computed, not answered)
| Rule | Checks |
|---|
| C1 Testing performed | Molecular testing done when required (NSCLC) |
| C2 Panel completeness | CGP, or all essential biomarkers individually tested |
| C3a Test ordered ≤ therapy | Testing ordered before/at first-line start |
| C3b Results ≤ therapy | Results available before first-line start |
| C4 PD-L1 testing | PD-L1 IHC performed for NSCLC |
| C5 Time to treatment | Diagnosis → first-line ≤ 60 days |
| C6 Targeted-therapy alignment | Actionable mutation got recommended first-line targeted therapy |
Out of scope: the experiment's benchmark harness, annotation tooling, and
scoring scripts — this package is the question + rule rubric only.
