OncoTwin
PT · 63F · NSCLC adenocarcinoma · stage III
connecting…
Doubling time— d
Chemo response—
Twin versionv0
Twin
Imaging
Compare
Cohort

Patient setup — drag to explore, the twin rebuilds live

Tumor volume · live

scan field
segmented tumor
day 0
42 cm³

Treatment strategy

Median @ day 365
—
90% interval
—

What's driving this forecast

day 0

Forecast trajectory

median forecast
90% interval
prior (pre-recalibration)
follow-up scan
Simulate a follow-up scan
Feed the twin a new measurement of this patient. It recalibrates its belief and the forecast snaps toward reality — the thing a static prediction can't do.

Scan · axial slice

scan slice

Segmented tumor

segmentation overlay

Radiomic features

pipeline: load scan → segment tumor → extract radiomics → fuse with clinical data → patient twin. The 3D shape in the Twin tab is this segmented mesh.

All treatment strategies · median trajectories

Projected outcome at 1 year

StrategyMedian (cm³)90% interval

Can we tell the strategies apart?

Model calibration (validation cohort)

What this shows

Backtested on an 80-patient synthetic cohort (assimilate 2 scans, forecast the rest). A well-calibrated 90% interval should contain the truth about 90% of the time. Prior-only and trained models were badly overconfident (~31–37% coverage); adding horizon-growing process noise brings coverage to ~90% without hurting accuracy. Reproduce on your machine with make analysis.

Real-data check · NSCLC-Radiomics

Tested against real patient survival (TCIA NSCLC-Radiomics, 399 patients): the twin's risk scored a concordance index of 0.505 — essentially chance. An honest negative result: the growth model was not trained on survival, and the prognostic signal lives in imaging features (tumor volume, heterogeneity) not yet loaded per patient. Real CT ingestion works; extracting tumor radiomics across the cohort is the next step. Reproduce with python analysis/real_data_analysis.py.