3 Ways AI Could Transform Clinical Trials
Artificial intelligence and Big Data could help solve many of the obstacles to effective clinical trials, including protocol design, data analysis, and patient recruitment and retention.
Artificial intelligence (AI) and Big Data have the potential to transform the clinical trial management process. These technologies, which already are speeding up drug discovery, could overcome many of the traditional challenges to effective trial execution, including protocol design and adherence, patient recruitment and retention, and data management and analysis.
Pharmaceutical companies spend billions of dollars every year on the clinical trial process, which is estimated to account for more than 40 percent of all pharma research budgets.1 And yet there are lost opportunities. According to CB Insights2, roughly 1.7 million people in the US are diagnosed with cancer for the first time each year, and there are more than 10,000 current clinical trials that could provide potentially lifesaving medicines. But less than 5 percent of patients get enrolled in trials (Opens Overlay). The use of AI to improve the structure and process of these trials may not only reduce costs and improve efficiency, but also lead to better patient outcomes.
1. Creating Smarter Protocol Design
Ultimately, the success or failure of a clinical trial often comes down to the design and execution of its protocols. A poorly structured trial can lead to excess cost, faulty data, difficulty in enrollment and significant delays. Clinicians often use prior design and results as a model to structure their future protocol efforts, but this can cause recruitment problems and extra costs if the historical data has flaws or if other pitfalls are not identified. AI can be used to cross-compare massive sets of data from past trials to find similarities and points of commonality, as well as any areas of concern that might need adjustment. This information can be used to better tailor the current trial. AI can also take operational performance from historical trials and predict the probabilities of success or failure in a protocol design, the likelihood of certain patient response rates, and even suitability or effectiveness of trial sites. In all, these AI tools are not meant to make decisions, but rather arm designers with better tools and information for structuring effective trials.
2. Improving Patient Recruitment and Retention
One of the greatest pitfalls in clinical trial design is effective patient recruitment. Even a properly structured and designed trial can be tripped up by ineffective or mismanaged recruitment of trial participants. According to a report on clinical trial costs, pharma companies spend billions annually on recruitment services for clinical trials, but despite this massive cost, roughly 80 percent of clinical trials fail to meet enrollment timelines and almost a third of phase III study terminations are due to enrollment difficulties.3 Just determining patient recruitment eligibility criteria for terminal illnesses (particularly with cancer) can be extremely difficult, as is connecting qualified patients with the right trials.
Instead of relying on a doctor to recognize that a patient is eligible for a trial, AI has the potential to tap patients’ electronic medical records to cross-reference hundreds of available trials based on specific criteria. The ability to sort and cross reference this massive data set would be next to impossible for doctors. It can also use existing patient information to assess the probability of a good fit for a given trial and ultimate responsiveness and success rates, thereby lowering dropout rates. Ultimately, this would lead to increased efficiency, more effective time estimates and lower costs.
3. Gathering and Sharing Data in Real Time
Another area where AI could greatly benefit clinical trial management is with the collection and analysis of data in the field. Typically, data is collected via clinical site visits and manual monitoring of responsiveness and results. The process often is extremely labor intensive, and the effort required to gather this information can lead to a poor patient experience, which ultimately affects adherence and dropout rates.
These problems can be greatly improved with the use of AI-managed telemedicine and digital applications, which can gather information remotely and in real time, share the information with doctors, and provide a two-way data stream for communication and trial management. Real-time data can also be analyzed immediately to find commonalities across symptom progression and outcomes, improving success rates.
The potential use of AI spans the clinical trial spectrum, and offers the promise of improving trial design, patient recruitment and data management. Understandably, much of the adoption and usefulness of AI in these processes will depend on more universal data protocols and the need for adherence to strict regulatory guidelines. This cannot be understated, and there is still a long way to go before these issues can be overcome and accepted. Still, AI could bring greater efficiency, reduced cost and positive patient experience to clinical trial management—and ultimately, more effective medicines for all of us.
In part three of our series on the role of AI and machine learning in life sciences, we will explore how the medical device industry is utilizing AI to revolutionize medicine.
1RDP Clinical Outsourcing, “Considerations For Improving Patient Recruitment Into Clinical Trials,” March 23, 2018
2A New York-based data firm, not associated with Commercial Banking or the Insights department of this site
3The Vector, “Healthcare and Artificial Intelligence: AI in Clinical Trials,” September 27, 2018