Interactive precision medicine revolution: unleashing a deep learning framework for drug response and recommendation

Abstract

Liver cancer ranks as the third most common cause of cancer-related death rate, resulting in an estimated 830,180 deaths globally. Recent, 2023 estimation of the American Cancer Society predicted around 29,380 deaths due to this disease in America alone. However, with early detection and proper drug management the survival rate can be greatly extended. Significantly, the noteworthy observation pertains to the absence of consistent drug responses among patients with identical cancer types at the same stage. Hence the need of the hour is the precision medicine that predict patient specific drug response based on molecular data genome expression. In this regard, we propose drug response system that uses deep learning framework that predicts drug response in new cell lines or patient with the aim of accurate prediction and suggesting précised medicine to that particular cancer patient. The prediction of the framework is on the lines of recommended system’s projection of drugs and cell lines into latent pharmacogenomic space. During our experimentation, we utilized the Genomics of Drug Sensitivity in Cancer dataset and observed that Cancer Drug Response and Patient Specific Recommender System (CDRP-PRS) consistently delivers robust predictions and effective response, even when applied to previously unseen patient-derived cell line datasets. CDRP-PRS inferences in the pharmacogenomic space are helpful to understand the drug mechanism to identify the cellular subtypes and their categorized associations. Further it is aimed is to predict the accuracy of drug responses for patients and suggest precise medications tailored to their specific cancer.

Keywords

• Cell lines
• cell clustering
• IC150
• genome expressions
• heat-map
• Pharmaceutical Science