Use of AI/ML and Digital Health tools in tuberculosis drug development - a feasibility study

The project herein involves evaluating whether AI and other digital tools can improve understanding of how to progress tuberculosis science to mitigate its global burden. Tuberculosis is the leading global infectious threat and the poor are disproportionately affected. With the increasing burden of HIV-tuberculosis co-infections, tuberculosis has become a high priority for research and action in global health. The goal of the treatment is to cure the patient, prevent death, prevent a recurrence, hinder the transmission of the infection and prevent antimicrobial resistance. The key to successful treatment of tuberculosis is adherence to medication which proves to be a significant challenge due to the long period of therapy, the use of polypharmacy and the presence of adverse drug reactions. GSK, a key partner in global efforts to combat TB, identified engagement with the University of Cambridge through this secondment as an efficient practice in response to the WHO's urgent call to save lives in poor settings and co-create impact. Endeavours to develop integrated and sustainable advances in tuberculosis treatment is a complex clinical, technological, social and economic challenge that could be resolved with artificial intelligence (AI) for big data analytics. The first objective of this project is answering a fundamental question in tuberculosis treatment: How to reduce the risk of nonadherence at a patient level? I will evaluate the use of AI platforms as diagnostic tools to derive in-depth insights from tuberculosis databases, identify adherence patterns and the benefit-to-cost ratio of interventions. This will potentially lead to recommendations for improving treatment adherence which, in turn, will reduce the tuberculosis burden, associated mortality, and financial consequences. The second objective is to accelerate drug development in tuberculosis through earlier identification of regimens most likely to succeed, thereby enabling prioritisation of the most promising regimens. In order to achieve this, I will capitalise on the predictive capacity of AI for sophisticated analysis of big data to evaluate the feasibility of designing and validating innovative clinical trials. I will assess the accuracy of AI-based designed clinical trials and simulations with respect to data quality and stability across populations and requirements for continuous optimisation. This disruptive innovation in clinical trials has the potential to increase the likelihood of success and sustainability of resources to identify novel treatment regimen(s), which leads to rein in growing research and development costs.