A Non-Interventional Pragmatic Clinical Trial of NLP Models for the Detection of Immune-Related Adverse Events to Improve Biorepository Registration

What this Trial Is About

Researchers are evaluating if an automated informatics system can improve the detection and registration of immune-related adverse events (irAEs) in patients receiving immunotherapy for cancer. The study aims to see whether this system is feasible for daily use and if it increases registration rates to a specialized irAE biorepository compared to standard monitoring methods. This trial is a randomized controlled study conducted at a single institution, focusing on patients with malignancy receiving immuno-oncology treatments. Participants will be randomly assigned to either standard irAE monitoring, where treating clinicians report eligible events, or to informatics-assisted monitoring, which includes a daily automated review of electronic health records (EHR) using the informatics system. The system runs behind the institution's secure firewall, identifying patients with new irAEs for potential registration in the biorepository. The study includes a feasibility run-in with the first 10-20 patients and plans to enroll about 100 participants in total. During the study, patients' EHRs will be reviewed to collect demographic, medical, and cancer treatment information. Participants will be monitored for irAEs for up to 12 months or until they register in the biorepository. Researchers will assess the feasibility of daily EHR analysis, registration rates, time to data entry after irAEs, and the accuracy of irAE detection based on expert review. The study will use a triple-masked randomized design to compare outcomes between the two monitoring methods.

Evaluating Informatics-assisted Immune-related Adverse Event Detection to Improve Registration Onto a Biorepository