Abstract OT3-33-01: Prospective, Multi-Center, Artificial Intelligence Study for Early Prediction of Serious Events under Treatment Is Now Open for Recruitment in Breast Cancer - OMCAT Trial in Progress

Abstract Background: Aim of the OMCAT trial (‘One Million CAncer Treatment months’, NCT04531995) is improvement of cancer patient care and safety by developing artificial intelligence (AI)-based, incident prediction algorithms. Incident detection allows early notification of treatment teams, enabling timely management changes or interventions. Ultimately the algorithms can also support improved health resource allocation. This trial in progress aims to provide learning databases in breast cancer comprising both electronic patient reported outcome (ePRO) data using the mobile medical device ‘CANKADO PRO-React’ and ground truth outcome data, which provide disease-specific events of interest (“incidents”) verified by the physician (e.g., during patient examinations). Methods: Incident prediction is posed as an application of stochastic time series analysis using AI and knowledge engineering technology. The learning process begins by fitting individualized and disease-specific stochastic process models to “incident-free” intervals extracted from the ePRO data series. Incidents produce detectable deviations from “ordinary” ePRO fluctuations. The algorithms are trained on CANKADO PRO-React data to produce real-time risk functions for predicting incidents on a clinically specified time horizon. Results: Considering the heterogeneity and combinatorics of diseases, stages, therapies, and types of events considered in this study, ultimately the AI algorithms aim to discover about 360 distinct predictive relationships. The estimate of one million treatment months is derived from statistical power analysis of this target, considering estimated median documentation time of six months per patient and estimated 400-500 patients per predictive relationship. To date, 45 centers in Germany have expressed interest in participating. This participation level will enable proof of principle. Ethics votes are already available in most regions. Other centers are invited to participate in this trial. Conclusions: OMCAT opens a whole new path towards evidence-trained AI and a novel combination of patient observation and predictive care. The goals of OMCAT are ambitious and will therefore require many more supporters. Citation Format: Timo Schinköthe, Christian Horst Tonk, Ronald Kates, Sherko Küemmel, Fatima Cardoso, Nadia Harbeck, Peter Staib, Annette Schmidt. Prospective, Multi-Center, Artificial Intelligence Study for Early Prediction of Serious Events under Treatment Is Now Open for Recruitment in Breast Cancer - OMCAT Trial in Progress [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr OT3-33-01.