Admin · Retrospective validation

Retrospective validation

Compare prototype scoring against baseline or reviewer-labeled benchmarks to assess ranking agreement, pathway consistency, and run stability. Internal validation only — not a clinical claim.
6 runs·Mean Δ +9.0 pts·1 in flight

Internal retrospective validation

These checks compare prototype scoring against baseline or reviewer-labeled benchmarks. They do not establish clinical validity or replace prospective trials. Use the results for internal review only.

Validation queue

New comparison run

Comparison

Lung adenocarcinoma

Baseline vs prototype relevance

How to read this chart

Baseline is the reviewer-labeled or rule-based ranking. Prototype is the OncoQ.tech relevance score. A positive delta indicates closer agreement on top-ranked mutation signals; a negative delta indicates the prototype diverges from the reviewer reference for that benchmark.

Top-k agreement:
How closely the prototype's top-ranked mutation signals match the reviewer or baseline top-10 list.
Pathway consistency:
How often the system maps mutations to the same pathway category as the benchmark.
Evidence agreement:
How often the evidence tier matches reviewer or baseline labels.
Run-to-run stability:
Whether repeated runs produce consistent ranking outputs.

Runs

Benchmark run log

6 results
IDLabelStartedStatusClassicalPrototypeΔReviewer
bench-0042Top-10 ranking agreement · LUAD retro4h agoComplete6271+9Dr. Tan Wei Lin
bench-0041Pathway mapping consistency · BRCA retro1d agoComplete7078+8Dr. Aiman Rashid
bench-0040Evidence-category agreement · CRC retro3d agoIn review6574+9Prof. Lim Boon Eng
bench-0039Top-k precision · LUAD retro6d agoComplete5869+11Dr. Tan Wei Lin
bench-0038Run-to-run stability · CRC retro8d agoComplete8492+8Prof. Lim Boon Eng
bench-0037Prototype layer noise sweep42m agoQueued————