A prospective study comparing highly qualified Molecular Tumor Boards with AI-powered software as a medical device.
Journal:
International journal of clinical oncology
PMID:
39714567
Abstract
BACKGROUND: The implementation of cancer precision medicine in Japan is deeply intertwined with insurance reimbursement policies and requires case-by-case reviews by Molecular Tumor Boards (MTBs), which impose considerable operational burdens on healthcare facilities. The extensive preparation and review times required by MTBs hinder their ability to efficiently assess comprehensive genomic profiling (CGP) test results. Despite attempts to optimize MTB operations, significant challenges remain. This study aims to evaluate the effectiveness of QA Commons, an artificial intelligence-driven system designed to improve treatment planning using CGP analysis. QA Commons utilizes a comprehensive knowledge base of drugs, regulatory approvals, and clinical trials linked to genetic biomarkers, thereby enabling the delivery of consistent and standardized treatment recommendations. Initial assessments revealed that the QA Commons' recommendations closely matched the ideal treatment recommendations (consensus annotations), outperforming the average results of MTBs at Cancer Genomic Medicine Core Hospitals.