Machine learning prediction of therapy response

A system comprising at least one hardware processor and a non-transitory computer-readable storage medium having stored thereon program code, the program code executable by the at least one hardware processor to:

• obtain, from each of a plurality of subjects, a biological sample;
• analyze the sample to identify at least one first single nucleotide polymorphism (SNP) associated with a trait of a specified type of a disease;
• calculate a first polygenic score based on the identified at least one first SNP;
• assess a statistical significance of the at least one first SNP with respect to said trait of the disease;
• select from the identified at least one first SNP one or more second SNPs that affect a response to a therapy for treating the disease, said selection based on:
• (a) location in genes that affect a target of said therapy, and (b) the assessed statistical significance;
• calculate a second polygenic score, based on the one or more second SNPs;
• and
• at an inference stage, apply a trained machine learning model to the first polygenic score and the second polygenic score pertaining to a target subject of the plurality of subjects, to predict a response of said target subject to at least one therapy from a set of therapies.