Error reduction in leukemia machine learning classification with conformal prediction

ABSTRACT Purpose Recent advances in machine learning (ML) have led to the development of classifiers that predict molecular subtypes of acute lymphoblastic leukemia (ALL) using RNA sequencing (RNA-seq) data. While these models have shown promising results, they often lack robust performance guarantees. The aim of this study was three-fold: to quantify the uncertainty of these classifiers; to provide prediction sets that control the false negative rate (FNR); and to perform implicit reduction by transforming incorrect predictions into uncertain predictions. Methods Conformal prediction is a distribution-agnostic framework for generating statistically calibrated prediction sets whose size reflects model uncertainty. In this study, we applied an extension called conformal risk control to ALLIUM, an RNA-seq ALL subtype classifier. Leveraging RNA-seq data from 1042 patient samples taken at diagnosis, we developed a multi-class conformal predictor, ALLCoP, which generates statistically guaranteed FNR-controlled prediction sets. Results ALLCoP was able to create prediction sets with specified FNR tolerances ranging from 7.5-30%. In a validation cohort, ALLCoP successfully reduced the FNR of the ALLIUM classifier from 8.95% to 3.5%. For cases whose subtype was not previously known, the use of ALLCoP was able to reduce the occurrence of empty predictions from 37% to 17%. Notably, up to 34% of the multiple-class prediction sets included the PAX5alt subtype, suggesting that increased prediction set size may reflect secondary aberrations and biological complexity, contributing to classifier uncertainty. Finally, ALLCoP was validated on two additional RNA-seq ALL subtype classifiers, ALLSorts and ALLCatchR. Conclusion Our results highlight the potential of conformal prediction in enhancing the use of oncological RNA-seq subtyping classifiers and also in uncovering additional molecular aberrations of potential clinical importance.