Objective
AI can be leveraged to speed patient identification and recruitment, reduce site burden, and empower patient engagement throughout clinical trials. It aids in identifying the right investigators and making the start-up process seamless will increase the probability of exceeding enrolment plans and completing clinical studies on time and on budget. One of the most important aspects of a trial is selecting high-functioning investigator sites. Site qualities such as administrative procedures, resource availability, clinicians with in-depth experience and understanding of the disease, can influence both study timelines and data quality and integrity. AI technologies can help biopharma companies identify target locations, qualified investigators, and priority candidates, as well as collect and collate evidence to satisfy regulators that the trial process complies with Good Clinical Practice requirements.
Outcome
AI would help to identify which sites would be able to provide us the required patient pool for the study and which would not be able to at an earlier stage. This will help in reducing the cost involved during the study start-up phase and would help in meeting the study timelines as per the stipulations. The solution helps to identify sites with the required capabilities, experience, highest enrolment, and leading start-up cycle-times.
Business Value
The solution can identify and mitigate potential risks and can be leveraged to identify the right investigators at the right time for the right study ( AI models can be used to develop the ideal target, investigators, and site lists for your specific indication and study protocol).ML can also be applied to the internal or external databases based on the number of patients the sites were able to provide and whether the inclusion and exclusions were similar, if not what were the differences, whether the site would be able to meet the new criteria, etc., This will give the figures in precession. Also using NLP, the written comments provided by any other CROs or team could be obtained as positive or negative feedback
H2O's AI and Data Approaches
This solution is powered by the H2O AI Cloud Driverless AI AutoML, H2O-3, and H2O.ai Wave. The data science approaches include genetic algorithm, advanced feature engineering, GLM, GBM, XGBoost, ensemble stacking, and Shapley value estimation, among others.