Automation, Analytics, AI and the Future of Trial Monitoring

High clinical trial failure rates often stem from difficulties in recruiting and retaining patients. A clinical trial may initially suffer from problems in finding and recruiting participants. And once they join the study, patients may drop out if the clinical trial team cannot monitor and coach them effectively.

High failure rates, in turn, drive inefficiencies in pharmaceutical research and development. Although investments continue to increase, fewer medications are brought to market when clinical trial failure rates stay high.

Artificial intelligence and automation can help clinical trial teams recruit and retain more patients. Predictive analytics allow clinical trial teams to make better predictions about patient behavior — so they can reach out before a patient vanishes from the study.

Challenges in Patient Recruiting and Retention

Finding and keeping patients in clinical trials pose two major hurdles for clinical trial teams. Recruiting patients can be difficult, especially when the drug to be studied targets a rare or unusual medical condition or combination of medical factors. For patients, factors such as travel, data gathering and side effects can become so onerous they choose not to complete the study.

Recruiting

Recruitment presents a challenge for every clinical trial team. “Matching the right trial with the right patient is a time-consuming and challenging process for both the clinical study team and the patient,” writes Anthony Lange, senior vice president of healthcare and life sciences at Virtusa.

The difficulty in recruiting shows in the numbers. Despite vast efforts in cancer medication research, only three percent of U.S. cancer patients are currently enrolled in a clinical trial, Lange notes.

Retention

Once patients agree to participate in a clinical trial, keeping them engaged poses additional hurdles.

An area in which patients often struggle is collecting the data required for a clinical trial. Tracking variables like medication use, treatment and side effects, vitals and other information can be overwhelming for patients.

In the face of the data-gathering burdens, many patients drop out.

Anticipating Patient Behavior

Currently, predictive analytics and AI can benefit clinical trial teams as they seek to understand patient behavior, writes Arun Bhatt in a 2021 article in Perspectives in Clinical Research. These include:

  • Machine learning (ML) and deep learning (DL), which can analyze large datasets to identify patterns and make connections.
  • Natural language processing (NLP), which can analyze written or spoken language to allow for more natural “communication” between humans and computers — and better analysis of written documents like electronic health records and medical research articles.
  • Human-machine interface (HMI), which helps humans communicate directly with devices. An AI-enabled chatbot that can respond directly to questions a human types into the chat window is one example of a human-machine interface.

As various analytics tools are combined and shaped toward specific goals, they develop more nuanced abilities. Predictive analytics, for example, uses machine learning to identify patterns that not only shed light on existing data but also indicate the probability of a certain event occurring in the future.

A 2022 study by Theresa N. Abiodun, Daniel Okunbor and Victor Chukwudi Osamor in Health and Technology examined the use of artificial intelligence and related tools in remote health monitoring. The resulting digital tools allowed clinical trial teams to classify clinical trial participants according to the likelihood that each patient would be able to continue with the study. The clinical trial team could then refer to the classifications as part of its evaluation of each patient’s continued participation in the study.

Currently, media attention on machine learning in healthcare tends to focus on healthcare delivery rather than on pharmaceutical research, write E. Hope Weissler and fellow researchers in a 2021 article in Trials. The lack of popular attention does not correlate with the level of investor and innovator interest, however. Efforts to apply machine learning to clinical trials continue to expand.

Emerging Frontiers in Automation for Clinical Trials

Artificial intelligence and automation open up new avenues for engaging with clinical trial patients.

“AI and ML methods may also be used to dynamically predict the risk of dropout for a specific patient, in other words to detect the onset of patient behavior that suggests the patient might be experiencing issues with adhering to the study protocol,” write Stefan Harrer, et al. in a 2019 article in Trends in Pharmacological Science.

For example, artificial intelligence and machine learning may allow a clinical trial team to spot severe side effects before a patient makes their own decision to discontinue a medication. AI and ML tools may also help ensure patients are taking the meds they claim to take.

A study by Earle E. Bain and fellow researchers examined patient compliance in a clinical trial studying medications for schizophrenia. All patients were told to return their empty blister packs so that the clinical trial team could track whether they’d actually taken the medication. In addition, patients were divided into groups: Some had their medication activities tracked via AI, while others were tracked on a non-AI computer system.

The authors found that the patients who received AI tracking were more likely to stay in the study and more likely to take their medication as prescribed. The AI-enabled system could more easily spot situations in which patients may not have taken their medication, such as when a patient evaded the view of the computer’s camera. These alerts allowed clinical staff to intervene in order to keep the patient on track.

AI may also help patients join and stay in clinical trials, write Nick Lingler and Siddharth Karia at Deloitte. For instance, AI may improve the overall clinical trial experience by:

  • Driving patient-centric trial design by offering new insights into existing patient data.
  • Optimizing the collection of data through sensors and wearable devices, and adjust in real time to improve each patient’s experience.
  • Reconciling various data modalities so that data can be analyzed together, regardless of its source.
  • Generating information needed to meet regulatory demands for certain documentation, freeing up clinical trial team members to focus on patients’ experiences.

AI and automation will continue to play a role in healthcare. As these tools develop, so will the need to monitor and adjust their use and performance. “Even after their successful integration into clinical practice, ML/AI algorithms should be continuously monitored and updated to ensure their long-term safety and effectiveness,” write Jean Feng and fellow researchers in an article in npj Digital Medicine.

Artificial intelligence, predictive analytics and automation increasingly offer ways for clinical trial teams to connect with patients and improve trial participation.

Learn more about our Data Science and eClinical Solutions.

Want to stay up to date with our news?

To top