Teaching AI to know when it is wrong about cancer immunotherapy, accepted at an IEEE conference

Immunotherapy works brilliantly for some cancer patients and not at all for others, and a model that predicts response is only safe if it also knows when it is likely to be wrong. Akhilesh evaluated exactly that across two real melanoma cohorts, 110 patients on CTLA-4 blockade and 143 on PD-1 blockade. He benchmarked three model families with leakage-free repeated cross-validation, measured calibration and patient-level uncertainty with bootstrap ensembles, and tested whether selective prediction and conformal prediction could flag unreliable calls. The result is an honest, rigorous answer to a question clinicians actually ask.

FIELDMedical AI & Oncology

RESULTFirst-author paper accepted at an IEEE conference

VENUEIEEE, 2026

BEFORE THE FELLOWSHIP

An 11th grader at Portola High School in Irvine, California with a background in R and Python and a strong interest in AI for medicine.

AFTER

First author of a clinically grounded study on immunotherapy response prediction and model uncertainty, accepted at an IEEE conference.

THE LEDGER

01First-author paper accepted at an IEEE conference

02Analyzed two real melanoma cohorts totaling 253 patients

03Error-detection ROC-AUC of 0.828 in the Van Allen cohort

04Leakage-free repeated five-fold cross-validation with ten repeats

05Accepted with minor revisions

NEXT CASE STUDYAmaay Gupta, First-author paper accepted at an IEEE conference

Every case study starts with one application.

ROLLING ADMISSIONS · SEATS LIMITED · NOW REVIEWING