Clinical trials play a critical role in advancing medical knowledge and patient care. They are designed to answer specific questions about biomedical interventions including new treatments (like novel vaccines, drugs, dietary choices, dietary supplements, and medical devices) and known interventions. They generate valuable data that contributes to scientific understanding of how our bodies work, diseases process, and health patterns.
One significant area of clinical trials focus is in HIV treatment research. Despite years of extensive research, the risk of failure in these clinical trials remains high. Failure, in this context, refers to the trials’ inability to end informatively - that is, their inability to sufficiently guide clinical, policy, or research decisions. Some uninformative trial outcomes are caused by inadequate design, such as underpowering.
We are using the concept of an informative trial proposed by Zarin et al in Harms From Uninformative Clinical Trials, JAMA 2019. You can read more about how the Clinical Trial Risk Tool quantifies the risk of a trial ending without delivering informative results in our article in Clinical Leader.
Recognising the high stakes and significant resources invested in these trials, Fast Data Science has developed a Clinical Trial Risk Tool that uses machine learning to predict the risk of a clinical trial ending uninformatively from its protocol text.
This tool leverages the power of advanced analytics to pore over the content of trial protocols, identify key parameters, and use these to identify problems with the trial design which could lead to a failure to deliver informative results. This algorithm-based analysis includes factors such as the number of participants, length of trial, and uncontrolled variables, among others.
As a result, the tool provides crucial insights that help researchers design more effective trials, assess risks, and predict their outcomes more accurately. For HIV clinical trials, this paves the way for more reliable and valuable results, informing better clinical and policy decisions, and bringing the world a step closer to reducing HIV infections as per the WHO’s 2022–2030 global health sector strategy on HIV.
While the algorithms aren’t perfect and the prediction isn’t absolute, the tool contributes greatly to the predictability and success potential of the trials. It offers a significantly enhanced understanding of the trials’ risk profiles, helping to optimise resource allocation and push towards more effective HIV treatment solutions.
HIV clinical trials carry inherent risks, but our AI technology can help identify problems at the design stage by analysing the protocol text. Researchers and clinicians can then make more informed decisions about the design, cost, complexity, informativeness, conduct, and interpretation of trials.
Find a clinical trial protocol on the US trials registry ClinicalTrials.gov.
You can then import that protocol example into our Tool to analyse for risk, cost, and informativeness.
