Clinical trials are the cornerstone of advancing medical knowledge and patient care. However, the risk of clinical trial failure looms large, and this is especially critical for cystic fibrosis, a rare inherited genetic condition that causes breathing and digestive problems.
Cystic Fibrosis (CF) trials come with a unique set of challenges. Trials are complex, time-consuming, and expensive. Recruitment is difficult. It is important that trials are conducted in a way that is likely to deliver informative results.
We are working with the definition of “informativeness” proposed by Zarin et al in Harms From Uninformative Clinical Trials, JAMA 2019.[1] 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.
A CF trial which fails to deliver informative results can have serious implications, both financially and, more importantly, for the patients who have been waiting for an effective treatment.
One of the significant issues affecting CF clinical trial outcomes is their informativeness - the ability of a trial to guide clinical, policy, or research decisions. A study by Hutchinson et al,[3] investigating trial informativeness, found only 26.4% of trials met four conditions for informativeness, adding a significant factor to the risk profile.
A pivotal question that arises is this: Is there a way to quantify the risk of an uninformative outcome associated with a CF clinical trial from its protocol?.
Fast Data Science’s Clinical Trial Risk Tool hopes to provide this estimate. The tool combines the lessons learned from past trials with state-of-the-art machine learning algorithms to predict trial risk from the protocol text.
This approach allows us to identify potential risks early in the trial design process and adjust accordingly, increasing the trial’s chances of informative completion.
The development of such a risk prediction tool becomes increasingly important given the cost and implications of CF clinical trials. Each trial, potentially costing more than £100,000 per patient per year, represents a significant investment.
Machine learning offers us the opportunity not just to cut costs and time in CF clinical trials, but to significantly increase our chances of success. It enables us to make better-informed decisions right at the start and control the risk parameters throughout.
