In 2012, Janey C. Peterson, Paul A. Pirraglia, Martin T. Wells, and Mary E. Charlson wrote that home visits during randomized controlled trials helped reduce patient attrition rates. Decentralized trials offer a technology-supported version of home visits. To date, finding the right technology to support decentralized trials proved challenging.
Developing technologies may help overcome these challenges, making tech-driven decentralized trials a feasible option in many clinical research settings.
Here, we take a look at emerging technologies that will facilitate the spread of decentralized clinical trials.
The Tech Challenges in Early Clinical Trial Decentralization
As technologies develop, clinical trial teams find ways to apply these tools to their work. Each new effort reveals shortcomings in existing technologies, as well as opportunities to improve these tools to build effective decentralized clinical trials.
Shortcomings in digital health user interfaces pose risks for decentralized trials, according to a report by the National Academies of Sciences, Engineering, and Medicine. When patients struggle to interact with a user interface, they may input data incorrectly, miss key health information, or even administer medication or other health aids improperly.
Another longstanding challenge in decentralized clinical trials is patient management of the data generated by the trial itself. In 2008, researchers Constance H. Fung and Ron D. Hays wrote that patients’ inability to manage a flood of data points hindered researchers’ ability to rely on patient-reported outcomes in clinical practice. The same is true for decentralized clinical trials. When patients can’t manage the data that their participation generates, that participation is of limited value to researchers — even if the patient doesn’t drop out of the trial.
Early uses of AI in patient interfaces focused on sending notifications when patients needed to complete a given task, such as testing blood sugar levels or inputting data. These efforts, however, were often hindered by the problem they attempted to solve. Patients could become so overwhelmed by notifications that they failed to complete the very tasks the notifications were intended to spur them to do.
Other challenges have also beset decentralized clinical trials. Privacy and security concerns, for instance, can hinder data collection, transmission, and analysis. New technologies make it easier to process patient-generated data securely in real time, making decentralized clinical trials more trustworthy sources of clinical trial data.
Artificial Intelligence and Decentralized Trials
Artificial intelligence offers a way to overcome several challenges that daunted earlier attempts to decentralize clinical trials.
In a 2022 article in Nature Medicine, Kevin A. Thomas and Łukasz Kidziński write that artificial intelligence offers a way to overcome two key challenges in clinical trial decentralization: creating user-friendly interfaces and managing trial data.
Incorporating AI into user-friendly interfaces can be done with improved automation, which can log and transmit data for patients. Another AI-based tool that can reduce the burden on patients is reinforcement learning, Thomas and Kidziński write.
Reinforcement learning, the technology behind computerized chess masters, can also be used to optimize notifications. The AI adapts to the patients’ interactions with the user interface, sending only the notifications the user needs. Rather than becoming overwhelming, notifications can fulfill their initial purpose: reminding patients to handle tasks only when the patient falls short.
AI can also help manage trial data. Automated data collection offers one example, but AI can adapt to a number of data-related tasks. For example, Kailas Vodrahalli and fellow researchers explore the ways in which AI can be used to improve the quality of photos used in telehealth appointments. Similar tools can help patients in decentralized clinical trials adjust their efforts in real time, improving the quality of collected data without unduly disrupting the patient’s routine.
Addressing Privacy and Security Issues
Privacy and security remain two top concerns both in the realm of digital data generation and in healthcare.
In a 2021 article in npj Digital Medicine, Walter De Brouwer and fellow researchers note that ensuring privacy and security in decentralized clinical trial technologies poses challenges. The researchers suggest that these challenges can be addressed by a combination of three technologies: edge computing, zero-trust environments, and federated computing.
Edge computing eschews server-based data processing in favor of data processing that occurs closer to the point at which the data is generated, such as a wearable medical device.
Edge computing showed early promise in reducing bandwidth and latency issues. As the technology is developed, it also shows promise in reducing overall data exposure, thus boosting security.
Edge computing continues to expand. Gartner’s Rob van der Meulen cites company research that predicts 75 percent of all data processing worldwide will take place outside traditional servers by 2025. Edge computing offers faster data processing for decentralized clinical trials plus the potential for better privacy and security surrounding protected health information.
A zero-trust security environment could also be called an “always-verify” environment.
In a zero-trust model, devices connected to the network are never assumed to be “safe,” no matter how many times they’ve connected before, writes Crowdstrike’s Kapil Raina. Rather, each connection point must demonstrate that it is what it says it is — every single time.
Wearable medical devices open up new vistas for decentralized clinical trials. Yet they also come with significant security risks.
“Existing vulnerabilities, poor configuration, and the use of default passwords are among the factors that can aid a hacker in compromising at least one device” connected to a network, writes Ziv Chang at Trend Micro Research. Since security in smart devices is often overlooked, smart devices are often the weakest points in any given network.
A zero-trust network environment compensates for any weaknesses in wearable security by requiring regular authentication and authorization. In so doing, the zero-trust network helps reduce the risk that a wearable device will become the source of a data leak.
Decentralized clinical trials also suffer from generated data sets that can’t be merged with one another. Data sets generated by individual collection methods or devices may require data entry or transformation by hand before they become interoperable.
Federated computing seeks to improve the interoperability of data. Rather than aggregating data sets in a central location for processing — and thus demanding that all data appear in the form the processing computations can read — federated computing moves the calculations to the data. Data is processed at or near its source.
Federated computing “paves the way toward enabling the processing of isolated data at scale,” writes Patricia Florissi, technical director for the office of the CTO at Google Cloud. Processing isolated data, such as information generated by wearables, at scale is essential for widespread adoption of decentralized clinical trials.
The COVID-19 pandemic accelerated the use of technologies to carry out essential tasks while maintaining physical distance between people.
Today, many of these technologies can be adapted to improve decentralized clinical trials. Artificial intelligence, zero-trust environments, and similar digital tools help clinical trial teams overcome previous hurdles to decentralized trial efficacy.
Images by: havucvp/©123RF.com, dragoljubbankovic/©123RF.com, milkos/©123RF.com