Artificial intelligence is continually evolving and expanding. Its use in medicine ranges from augmenting human judgment to full task automation, and it has potential to transform clinical research. In a JAMA editorial, Dr Adrian Hernandez and Dr Christopher Lindsell from Duke University in North Carolina, United States, explored the potential role of AI in the future of clinical trials.
Sponsors of clinical trials face many obstacles, including high financial cost, lengthy time frames, difficulties in recruitment and retention of participants, and a disconnect between clinical results and real-world outcomes. Global expenditures on clinical trials are significant (>$50 billion) and are estimated to surpass $85 billion by 2030. Up to 20% of these costs is related to manual transfer and verification of data from the electronic medical record to a data capture system, as well as site monitoring of such data. These are readily automated tasks that could potentially be revolutionized with the application of AI.
Natural language processing models may be able to accurately identify clinical events and improve the efficiencies of multicenter clinical trials, yielding significant cost and time savings. AI could potentially enhance trial recruitment by mining electronic health records to identify prospective participants and interacting with clinicians and patients to help them find studies that could be a good match. AI might also be used to translate a protocol into a manual of procedures and schedules of events, with customized data-collection tools to support the integration of research into patient care. Another potential use is for dissemination of customized information through local, regional, and health care communities.
To achieve these benefits, the authors suggest that rigorous validation and regulatory oversight are essential to ensure safe, effective, and ethical deployment of AI in the clinical trials ecosystem. They propose 4 key domains that should be included in a regulatory framework: reliability and validity, transparency, generalizability and bias, and privacy.
High level
AI offers enormous potential, but a rigorous framework is critical to evaluate the accuracy of AI approaches so that it can be safely used in clinical trials for oncology or other disease states. Oversight will be needed to ensure that historical biases that have led to inequitable study populations in the past are not carried forward as new models are developed. Researchers will need to collaborate and share their findings to generate high-quality evidence and advance this technology into the future of clinical trials.
Ground level
The science behind AI-augmented clinical trials is building, and it has enormous potential to increase efficiencies and reduce inequities. Clinicians may be able to use AI to identify appropriate clinical trials for their patients. For study participants, the future of clinical trials could one day include the use of AI to conduct virtual follow-up visits and assess patient-reported outcomes, thus transforming traditional approaches of evaluation.
