How Artificial Intelligence is revolutionising patient compliance

As a Senior Director of Clinical Science, William Jacobson is responsible for driving strategy for global clinical development at Takeda. At T3: Trials, Tech and Transformation Jacobson is speaking on the intriguing topic of how Artificial Intelligence can be used to improve subject compliance. We caught up with him on the tricky issue of compliance and how technology is helping meet the challenges.

What are the current issues with subject compliance during clinical trials?

WJ: ‘Compliance with investigational product use in a clinical trial is essential if one is to determine the true safety and efficacy profile of that product, whether in comparison to placebo or an active comparator. Typical means of compliance, historically, have been self-report and pill count. However, these have repeatedly been demonstrated to be unreliable, frequently overestimating the degree of compliance - pill counts for example routinely indicate levels of adherence above 95% across therapeutic areas.

Over the years a number of methodologies have been developed to increase our accuracy in determining compliance, as well as the level of compliance in clinical trials. In general though, they have not been particularly successful, with compliance often not determined until the end of the trial, at which point it is far too late to intervene in a positive manner.

For example, an alternative to pill count is the measurement of concentrations of a drug or its metabolite. However, between-person variability in pharmacokinetics (PK) as well as limits on the frequency of blood sampling prevent this approach from providing accurate day-to-day information on adherence.

The advent of technology over the last few years has certainly advanced this, but both accurate determination of, and subsequent intervention to improve compliance still remain problematic. This is clearly an issue. When an investigational drug fails to demonstrate a beneficial effect in a clinical trial, it is difficult to determine whether the drug is truly ineffective or it worked sub-optimally because subjects were non-adherent. Non-adherence results in reduced signal level and diminishes our ability to detect an effect.’

How does the Artificial Intelligence platform work? How does it meet these challenges?

WJ: ‘The clinically-validated AI platform uses software algorithms on smartphones to visually and automatically confirm participant identity, the medication, and medication ingestion. Participants receive automated reminders and precise dosing instructions. The AI platform also offers facial recognition in combination with ePRO to ensure accurate data collection of all patient-reported outcomes. Study teams have access to real-time data for review and intervention. Early detection of non-adherence allows for immediate follow up and ensures high completion rates and high rates of adherence.

Continuous and virtual monitoring of adherence in ambulatory settings allows clinical trials to ensure the same quality of data achieved for inpatient participants and automates the process of medication reconciliation. In addition, the impact of the platform on adherence and drug concentration translates into higher statistical power resulting in cheaper, faster and more successful trials.’

What other novel digital and mobile compliance technologies are you most excited about?

WJ: ‘There are two other novel compliance technologies I focus on. The first is from Proteus Digital Health. It consists of an ingestible sensor, a wearable sensor patch, an app that works on a mobile device and a portal through which investigators or sponsors can access data. The ingestible sensor is incorporated into the investigational product.

Secondly, there is the Xhale Smart Medication Adherence System. This technology requires the incorporation of an additive or adherence marker into the investigational product formulation. It provides a cue to remind the subject to take the medication and after ingestion the subject blows into what is in effect a portable gas chromatograph. The device detects the presence of a metabolite of the adherence marker and confirms compliance with IP ingestion. The data is time stamped and uploaded to a database.’