INFLUENCE 3.0 was derived from 36,000+ Dutch patients and has never been validated on Indian cohorts. INDIRA re-derives the model on Indian-representative data with formal competing-risk correction.
~25% Indian-representative, versus ~12–15% in INFLUENCE 3.0.
Weighted for NCGEQAS pathology concordance across Indian labs.
Fine-Gray models from Phase 3 — not formally corrected in the benchmark.
Indian cohort literature IPD plus a SEER proxy, not Dutch registry data.
Published figures and AlgoMD research targets. Targets are goals, not results.
Indian breast cancer presents a decade younger, with higher TNBC and HER2-positive rates, more advanced staging and different therapy access. A Dutch-derived model cannot produce calibrated estimates for these patients.
Mean presentation age in India, versus 62–64 in Western countries.
Triple-negative rates roughly double those of Western cohorts.
Patients behind INFLUENCE 3.0 — none of them Indian.
National Cancer Grid institutions — the network INDIRA is designed around.
INDIRA is in active data acquisition, combining registry extraction with individual patient data from published Indian cohorts.
SEER 21 breast-cancer data for proxy training — extraction in progress.
Seven published Indian cohorts digitised via Engauge and IPDfromKM.
Registration for controlled-access Indian breast-cancer genomics data.
National LRR incidence data via the Cancer Samiksha portal.