INNV-16. ONCOPATH: OPTIMIZED TREATMENT GUIDELINE CONCORDANT PATHWAY PRESENTED ON A VISUAL-INTERACTIVE DASHBOARD FOR CLINICAL DECISION SUPPORT COMPUTED BY OVERLAYING PATIENT LONGITUDINAL DATA ON A KNOWLEDGE GRAPH

Le résumé fourni par la source

Abstract BACKGROUND OncoPath provides a visual analysis of a brain tumor patient’s longitudinal clinical data overlayed on disease specific pathways with the goal of reducing knowledge discordant care and insurance authorization burden. By ingesting, curating and visually presenting the patient experience on guidelines, OncoPath aims to streamline clinical decision making and related processes. Understanding the patient’s journey compared to treatment guidelines is of value in addressing health equity and guideline adoption in real world settings. METHODS Data from 44 glioma patients diagnosed and treated between 2016-2021 were uploaded to OncoPath using natural language processing and other tools to capture abstractable data elements. The data was overlayed on guidelines using recursive graph modeling. Using the knowledge graph of a patient’s history, the model also recommends treatment options in an interactive visual dashboard representing NCCN guidelines. The dashboard includes the guidelines in graphical format with associated references and notation. RESULTS 28 males and 16 females age 21-38 years at diagnosis were abstracted. Cases represented 4 oligodendrogliomas, 13 astrocytomas, and 27 glioblastomas. Data was available through second line therapy, discharge to hospice or death. Cases were matched to the NCCN 2021 guidelines which was used for treatment decisions until November 2022. The patient data matched OncoPath except in 3 cases where KPS was not available resulting in premature pathway truncation. For these cases we inferred KPS based on subsequent treatment received to optimize the historic data. CONCLUSION To our knowledge, this is a first-of-a-kind technology in neuro-oncology that may improve time to treatment, reduce health utilization resources and can serve as a benchmarking tool for care delivery. The feasibility of clinically implementing such tools for decision support was demonstrated. This type of tool could be particularly useful in low-resource areas where disease specific expertise may not be available or to illuminate care discrepancies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.