One of the significant challenges we face in the fight against COVID-19 is the rapid development and testing of potential vaccines and treatments. Clinical trials carry significant risks of failure, both in terms of interventions that do not work, and trials which are poorly designed and are unable to deliver informative results.
A failure to deliver informative results means that the trial may not provide the required answers to guide clinical, policy, or research decisions. Primary factors such as the number of participants (is the trial suitably powered?), the duration of the trial, and the anticipated effect size, can contribute to a trial being at risk of ending uninformatively. For example, if a Statistical Analysis Plan (SAP) has not been drafted along with the protocol, we would consider the trial to be at risk of ending without delivering informative results.
We are using the concept of “informativeness” which was 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.
For COVID-19, the problem of uninformative trials was particularly acute due to the rapid speed of development of the first COVID-19 vaccines during the pandemic. It became clear during the nationwide and worldwide vaccine rollouts, that some vaccines failed to live up to the promised effectiveness.
According to a study by Hutchinson et al.,[1] only 26.4% out of 125 examined clinical trials met the conditions for informativeness. Given the urgent need to find effective remedies for COVID-19, such a high rate of failure could lead to a significant waste of resources and time. Therefore, it becomes crucial to predict the potential risk of a trial beforehand using methods like machine learning, which brings us to Fast Data Science’s Clinical Trial Risk Tool.
Fast Data Science’s Clinical Trial Risk Tool uses advanced machine learning techniques to predict the risk of a clinical trial based on the protocol text. The protocol text, outlining the trial’s methodology, is fed to the tool to analyse the parameters affecting the trial’s informativeness. Learning from previous successful and unsuccessful trials, the tool predicts the potential risk of an uninformative outcome associated with a new COVID-19 clinical trial.
With the surge in the number of clinical trials initiated for COVID-19 interventions, this tool can play a pivotal role in ensuring good trial design. By providing an estimation of the trial risk upfront, the tool allows researchers to make informed decisions, tweak the trial design if necessary, and thereby increase the trial’s chance of ending with informative results.
