Patient attrition can be a major barrier to the successful execution of a randomized controlled trial (RCT). When patients drop out of a trial after enrollment or randomization, not only does it impact the results, but it also introduces delay and higher costs. Today, with the help of artificial intelligence (AI), clinical study managers have access to the data they need to proactively reduce attrition rates in RCTs.
Medication Non-adherence as an Early Warning Sign of Attrition
In clinical research, artificial intelligence is playing an increasingly important role. In addition to helping uncover insights that were previously hidden in large datasets, computer vision and machine learning are being used to collect prospective data about patient attrition. This type of data can provide invaluable information on whether or not patients are planning to take the study drug as instructed—a leading indicator of whether they will drop out. Furthermore, understanding the precise reason for patient attrition can help study managers leverage innovative strategies that may improve patient retention.
Predictive Medication Adherence can Help Reduce Attrition Costs
A new benefit of leveraging AI in clinical research is that it reduces manual effort and cost associated with patient retention. If a patient believes they’re taking a placebo or that they are experiencing some adverse effect, it’s quite likely that they will drop out.
AiCure’s platform enables researchers to remotely assess dosing behavior using smartphones, computer vision and machine learning. Patient behavior can trigger customized and automated reminders and notifications when a person has missed a scheduled appointment or medication dose. This allows for quick action and intervention if necessary, which helps prevent potential dropouts from occurring in the first place.
AiCure Patient Connect gives sites a unique view into patients who don’t take their medications as directed. But what’s more important is it gives clinicians the ability to understand why patients stopped taking the treatment by actually talking to them through site-enabled secure calls, text, and video chats.
AiCure algorithms can also predict patient adherence and help reduce both the costs of attrition and the extra recruiting needed to meet enrollment targets. For example, in an early phase study of a central nervous system disorder, AiCure worked with the study sponsor to develop predictive models of drug non-adherence that successfully predicted high-risk patients. The models can also help identify patients who are generally adherent, those who are adherent with continued support, and those who are intentionally non-adherent.
Additionally, AI can be used to analyze large datasets related to patient engagement and satisfaction surveys which can help identify potential causes for dropout before they occur. Patient engagement and satisfaction, when coupled with dosing data, provide deeper insights about patient risks. As an example, AiCure developed a risk assessment tool in an innovative study of HIV oral pre-exposure prophylaxis (PrEP) by combining medication adherence and behavioral data using electronic patient reported outcomes (ePRO) technology.1
Patient Attrition in Clinical Trials Is Troubling But Preventable
Attrition in randomized control trials is a troubling yet preventable part of clinical research. Patient dropout often triggers additional recruitment efforts, leading to substantial costs and delay. Over the next several months we will explore the impact of patient attrition across a variety of therapeutic areas and the opportunities for improvement using predictive adherence.
Key to the reduction in dropout is artificial intelligence that provides sponsors with access to prospective data about why patients may drop out so they can take proactive steps to reduce attrition rates and improve outcomes. With the ability of AI to automate processes and quickly analyze large datasets related to patient engagement and satisfaction, computer vision and machine learning in clinical research have become invaluable tools for optimizing RCTs.
1Buchbinder, S.P., Siegler, A.J., Coleman, K. et al. Randomized Controlled Trial of Automated Directly Observed Therapy for Measurement and Support of PrEP Adherence Among Young Men Who have Sex with Men. AIDS Behav 27, 719–732 (2023).
