On average, the medicinal drugs completing their fourth and final phase of clinical research studies this year first entered the clinical trial process the same year the global recession was catalyzed by the bankruptcy of Lehman Brothers, or as some of us are more likely to remember, the year High School Musical 3: Senior Year was released in theaters. That was ten years ago.
Most drugs in the United States undergo ten to fifteen years of clinical testing before they are approved by the Food and Drug Administration (FDA). Usually, clinical trials involve four phases designed to test the efficacy of the drug in human populations. Clinical trials are also a notoriously time-consuming process, often involving the mobilization of large sums of monetary investment. Specifically, some estimates put the clinical trials market as worth almost 65 billion dollars. The laborious process of clinical trials, while designed to ensure the safety of a drug before allowing it to enter the market, can also inadvertently delay a patient’s access to potentially life-saving medication. Therefore, the clinical trials market is a space in need of a streamlined overhaul. Into this space enters the seemingly boundless potential of artificial intelligence (AI).
Much has been reported regarding the diagnostic potential of AI to assist physicians with parsing massive amounts of data, including historical presentation of symptoms of different diseases from thousands of patients. However, AI can also prove to be an advantageous addition to the clinical trials process. Given that clinical trials include several phases, preceded by a massive mobilization of matched volunteers, there are several spaces in which AI can be integrated to streamline the process.
Part of the delay and costs associated with drug research and development is identifying and matching volunteers for each phase of the clinical trial. The need for a more effective matching system is evident, as a report from Cognizant reveals that approximately 80% of clinical trials are unable to meet enrolment deadlines. In this space, AI can prove enormously helpful, as the algorithmic nature of machine learning can easily process copious collections of medical records, and sort and match those patients best suited to particular clinical trials. The magnitude of the impact AI can have on this step of the clinical trial process cannot be understated, as in the United States there are almost 20,000 clinical studies recruiting patients presently.
AI can also serve as a novel method of measurement of treatment outcomes. A case study for such integration of AI is the use of Orbita.ai and the Amazon VOICE system to build a voice activated questionnaire and engagement feature for potential use in clinical trials. Specifically, the AI questionnaire and engagement feature was used in a test trial with healthy volunteers to assess its effectiveness as a measurement tool for a patient’s condition. The trial revealed that the strength of AI’s use in clinical trials is that it allowed ease of volunteer participation by transcending the rigidity of a mere app experience, instead providing engaging conversation that stimulated volunteer engagement. Engagement is of paramount importance in clinical trials, as volunteers actively participating in the trial are more likely to exhibit retention and regular participation. Furthermore, the trial with Orbita.ai found that 70% of volunteers recruited for the trial preferred interacting with the AI interface over app entry or voice input alone.
Although the Orbita.ai trial proves promising, the introduction of AI into any phase of the clinical study process in drug development must be approached with caution. While the potential of AI to streamline the process of identifying and matching patients before beginning of clinical trials, as well as assist with data collection during clinical trials is formidable, the issue of security remains. Assurance of the security of patient data remains the rate-limiting step of AI integration into clinical trials.