Knowledge discovery and machine learning to elucidate the mechanisms of HIV activity and interaction with substance use disorder

项目总结

PROJECT SUMMARY More than 36 million people worldwide are estimated to be living with HIV infection and more than 1.2 million are in the USA. With the introduction of highly active anti-retroviral therapy, the life span of HIV-infected individuals has increased significantly. However, the quality of life of can be compromised owing to a range of cognitive deficits and memory loss, commonly referred to as HIV-associated neurological disorders (HAND). HIV-infected individuals are more likely to suffer from substance use disorder (SUD), and disproportionately suffer from high all-cause mortality. Drugs of abuse also increase severity of HAND by several potential biological mechanisms. HIV associated cognitive deficiencies in conjunction with SUD decrease engagement in HIV care, which fuels a worsening downward spiral of health status. Despite intensive research, there is no approved therapy for the treatment of HAND and particularly for the combined neurological effects of HIV and drugs of abuse. We have developed and employed MOLIERE and AGATHA, AI-based literature mining systems that discover novel interactions that potentially contribute to HAND. These systems also prioritize mining results to uncover small molecules that can be tested for anti-HAND therapy. Experimental validation of MOLIERE was achieved; four small molecules predicted by MOLIERE were shown to prevent HIV-Tat and cocaine induced neurotoxicity. AGATHA improved MOLIERE results on a massive retroactive validation and is ready to be deployed for wider searches that now include PubChem. In parallel, our previous efforts querying the Department of Veterans Affairs / Veterans Informatics Network Computing Infrastructure (VINCI) with specific hypotheses have successfully uncovered potential associations of unanticipated modifiers of HIV-associated pathologies. Collectively, these results led us to the central goal of this proposal to develop and apply an integrative AI-based approach to analyze biomedical datasets and Electronic Health Records to determine new mechanisms of HIV and substanses of abuse interactions, and to discover repurposed drug candidates to be tested for the treatment of HIV-infected SUD patients. This will be accomplished in three Aims. Aim 1 will develop a multidimensional AI-based text mining approach to explore new mechanistic connections between HAND and substanses of abuse. This will generate new knowledge of HAND and SUD interactions, and uncover small molecule and drug candidates that can be tested for activity against the neurotoxic insults caused by HIV and substanses of abuse. Aim 2 will develop and apply advanced machine learning and AI algorithms to explore health records of HIV and SUD patients. The outcome will be the development of the machine learning system to analyze VA data and generate of signals (hypotheses) for medications or medication targets that might have value to experimentally test for repurposing to manage HAND. Aim 3 will prioritize the selected candidates for experimental validation and further clinical development.