Age of AI

Transforming risk-based monitoring with real-world data

Year after year, we watch as the clinical trial landscape shifts. Making adjustments for speed, safety and patient-centricity is important, and conducting trials that mitigate risk is crucial for delivering market-ready therapeutics.

But when it comes to risk and to risk-based monitoring (RBM), there’s always room for more innovation. RBM is a fundamental part of clinical trials. Indeed, risk to patients and to the clinical trial process in operations are the key focuses of prevention, reduction and mitigation.

Even though the tools and technologies that enable centralised and remote monitoring have existed for some time now, monitoring tended to take place on-site.

COVID-19 forced a change that was already brewing in the industry, with 94% of monitoring taking place remotely in April 2020, compared to just 20% in the January of that year. Not only did it reduce disruption, but analysis showed that the ‘total non-COVID protocol deviations detected each month from March to May were similar to the February baseline’.

In other words, researchers could achieve the same standards of data quality and patient safety when monitoring off-site.

While there is still room for further adoption, the availability of real-world data (RWD) from wearables and other devices is helping to inform, and improve the effectiveness of RBM. RWD investments throughout the industry are being leveraged more frequently to support clinical trial feasibility assessments through various activities.

Some of these activities include identifying more precise target patient populations, hypothesis testing, safety signal predictions and informing diversity, equity and inclusion plans and performance, to name a few.

Advanced analytics, driven by AI and machine learning (ML), are the backbone of this process, making it possible for researchers to interpret large amounts of data and proactively identify potential risks to patients and/or processes. When multiple global trials are running concurrently, it allows risk factors to be continually reassessed based on real-time information.

AI also enables decisions linked to RBM to be automated, while keeping humans in the loop, helping to make trials more efficient and potentially detect potential systemic biases in the data, giving researchers an opportunity to assess and reduce errors.

Predictive analytics, meanwhile, help to reduce uncertainty and mitigate risks to patient safety and trial performance.

This exchange of data can be bidirectional, with RWD being captured through RBM to develop a deeper understanding of how adherent patients may be to a clinical trial protocol.

For RBM to be more widely adopted, researchers need assurances that they can manage data in a secure and regulatory-grade environment, where it is cleansed, standardised and interrogated throughout the trial life cycle.

Transparency is critical. Researchers must be able to identify any inaccuracies, duplicates and gaps in the data quickly, and integrate data from different sources with ease and without error. With quality data, researchers can then develop clear and repeatable processes that support the required ongoing monitoring during a clinical trial.

The application of AI and ML to RWD helps to speed up risk monitoring and identify patterns and trends.

Without proper oversight and human accountability, there is a danger of bias in AI-enabled decisions, meaning that differences in risk may not be identified among particular groups of participants.

Researchers must be able to rely on their analytics partner to continually validate, monitor, test, update and redeploy their models to ensure equitable and ethical patient-centric outcomes.

RWD and RBM are a powerful force, which can vastly improve clinical trials, but they are nothing without trust and transparency.

Sherrine Eid is Global Head, RWE, Epidemiology and Observational Research at SAS. Go to sas.com