Prototypical Visualization of Patient Similarities in cBioPortal to Enhance Decision-Making in Molecular Tumor Boards

Abstract

Background: Patient similarity analysis is pivotal in cancer research and clinical oncology, aiding in identifying patterns among patients with similar clinical and molecular profiles to guide therapeutic decisions, particularly in Molecular Tumor Boards (MTB), where therapy decisions are frequently informed by the treatment experiences of previously treated similar patients. However, the lack of standardized tools for automation and visualization limits efficiency here, especially in individualized MTB decisions. Objective: This study aims to develop a graphical user interface that aligns with clinician preferences to enhance patient similarity assessments and support decision-making in MTBs. Methods: Visualization concepts were developed through iterative design and evaluation cycles involving clinical experts. Mock-ups were created to represent various approaches for displaying patient similarities, focusing on molecular data relevant to MTB decisions. Results: Various designs were developed for visualizing patient similarity in cBioPortal. These include tabular views, network representations, and radar plots. Conclusions: These visualizations offer promise in enhancing decision-making in MTBs by making patient similarity assessments more accessible. Future development will focus on additional functionalities and better integration into clinical workflows.