Predictive Biomarker Profiles in Cancer Using a Unique AI Model Based on Set Theory

Abstract

The current study applies a new artificial intelligence (AI) method, ALiX, which is based on interval arithmetic, to analyze and interpret biological data for a clinical problem: identification of biomarkers for cancer diagnosis. Key unique and important features of this study is that ALiX provides an explanation to our medical hypothesis in the form of a list of ranked protein biomarkers that identifies which biomarkers are the most significant drivers of the predicted outcome, a capability that is not currently available in other AI applications. This study identifies a unique profile for stratifying cancer patients and for further stratifying the patients with cancer into subtypes that respond to treatment or not.