Trial prospector: An automated clinical trials eligibility matching program.

Abstract

6538 Background: Clinical trials are the evidence base for improving cancer treatment. Unfortunately, only approximately 5% of cancer patients (pts) take part in clinical research studies. Even in settings where clinical trials (CTs) are available, pt participation remains low. We hypothesize that the time required to identify appropriate studies for individual pts is a significant barrier to clinical trial accrual. Methods: Our multidisciplinary team developed Trial Prospector (TP), an innovative and flexible computer-based system that utilizes artificial intelligence and natural language processing to automatically extract information (e.g. demographics, pathologic diagnosis, stage, labs) from multiple clinical data systems and then match it to CT eligibility criteria. A user-friendly interface allows physician perusal of relevant CTs and eligibility checklists at the point of care without requiring manual data entry. We pilot tested TP in our cancer center GI oncology subspecialty clinics. TP was deployed for consecutive new pts, and oncologists (oncs) completed surveys after each visit to assess the usability and impact of TP. Results: Eleven medical oncologists (6 attendings and 5 fellows) participated in the pilot study. TP was deployed during 60 new pt visits. Of the 15 relevant GI/phase I CTs, TP identified a mean of 7 ± 2.7 eligible trials per patient. The most common reasons for ineligibility were pathologic diagnosis and labs. CTs were considered by the treating onc for 66.7% of the pts. 95% of participating oncs reviewed the TP output at the point of care with 70% spending 0-5 minutes assessing eligibility. Of the pts considered for CTs, a TP report was reviewed 72.5% of the time. Oncs reported that TP saved time identifying potential CTs during 57.1% of the visits. The reports were manually reviewed, and the TP matching algorithm was 100% accurate. 90.9% of the oncs recommended TP for CT screening. Conclusions: These results indicate that Trial Prospector is a feasible, accurate, and effective means to identify CTs for individual pts during a busy outpatient oncology clinic. Ongoing refinements will expand clinical data extraction and CT warehouse to improve precision and applicability across diseases.

Ähnliche Arbeiten