AI tech TrialGPT can cut screening time by 40%, says NIH

An artificial intelligence (AI) algorithm developed by the US National Institutes of Health (NIH) successfully matched patients to clinical trials faster than traditional approaches, according to a new study.

The technology, called TrialGPT, was created by National Library of Medicine and National Cancer Institute researchers who combined several “large language models” – AI systems that read text – into a platform able to automatically match patients to studies.

TrialGPT uses a four-step process, the first of which is a detailed examination of patient information, including relevant medical and demographic data. This is followed by a “retrieval” step where the system identifies relevant clinical trials by parsing the ClinicalTrials.gov website.

The third step, called TrialGPT-Matching, looks at the eligibility of each patient for each study. This is followed by TrialGPT-Ranking where the system uses the matching results to generate a ranked list of trials based on the eligibility of a given patient.

Performance

Based on a study that tested TrialGPT on 183 “synthetic patients” with over 75,000 trial annotations, the approach shows promise.

According to the results, which were published in Nature Communications this month, the system achieved a patient-to-trial matching accuracy of 87.3%, which is “close to expert performance.”

In addition, the ranking scores were “highly correlated” with human judgements and outperformed the best-competing models by 43.8%. The study also showed TrialGPT reduced screening time by an average of 42.6%.

Lead author and NLM senior investigator Zhiyong Lu said, “Our study shows that TrialGPT could help clinicians connect their patients to clinical trial opportunities more efficiently and save precious time that can be better spent on harder tasks that require human expertise.”

The findings were welcomed by NLM acting director Stephen Sherry, who said, “Machine learning and AI technology have held promise in matching patients with clinical trials, but their practical application across diverse populations still needed exploration.

“This study shows we can responsibly leverage AI technology so physicians can connect their patients to a relevant clinical trial that may be of interest to them with even more speed and efficiency.”

DepositPhotos/AntonMatyukha