Facilitating precision medicine in the community oncology setting: Artificial intelligence clinical decision support and expert consultation combined in a demonstration project.

Abstract

e13693 Background: More than 75% of cancer patients are treated in the community oncology setting. Data show that the majority of patients (~70%) eligible for next generation sequencing (NGS) do not get NGS testing, thereby jeopardizing improved outcomes. The purpose of this demonstration project was to determine the extent to which an intervention that combined a patented Artificial Intelligence (AI) clinical decision support (CSD) platform with expert consultation: 1) improved the identification of potentially eligible patients for NGS testing; and 2) facilitate therapeutic decision-making. Methods: A demonstration project, lasting approximately two months, was designed to improve uptake of NGS testing in a community oncology network. Baseline data on NGS testing rates were gathered prior to the introduction of the intervention. Education regarding the purpose of the intervention and implementation plan for the protocol was provided to participating oncologists. The intervention was introduced following the collection of the baseline data and educational session. Platform data were used as the source of data processed by the AI inference engine to: a) identify potentially eligible patients for NGS testing; b) facilitate automated ordering of NGS tests; and c) facilitate access to expert opinion. Data from the platform were also extracted to determine what therapeutic decisions were made, and the extent to which any of the therapeutic decisions had been modified in response to the CDS and expert opinion feedback. Results: 50 consecutive advanced cancer patients receiving NGS testing per guidelines, were assessed as part of the project (21 NSCLC, 11 CRC, 8 Pancreas, 4 Breast, 2 Biliary, 2 Bladder, 2 Prostate). All NGS tests were ordered via platform automation. The testing took 16 days (range 11-28 days), with expert consultation adding 9 hours (range 1-28 hours). All patients had NGS results that impacted therapeutic options, including 42% with ≥ 2 therapy matches and 22% with options beyond the NGS report. Over 30% of treatment decisions were altered based on CDS AI+expert review. Conclusions: The combination of AI clinical decision support and expert opinion facilitated the identification of cancer patients eligible for NGS testing and improved therapy decision-making. A larger study with longer-term assessment of the outcomes of these decisions is being planned.