Data science and epidemiology to improve health of people with HIV - PROJECT SUMMARY / ABSTRACT Many people with HIV (PWH) are living to older ages with the availability of efficacious and tolerable antiretroviral therapy, especially when diagnosed and linked to care early after infection. But healthy life expectancy among PWH is limited by comorbidities, including mental health, substance use, and age-related conditions affecting PWH at younger ages. Hospitalizations among PWH merit study as these sentinel events are valuable for understanding disease burden, trends in morbidity, health disparities, and healthcare utilization and needs. Hospitalizations are also a sensitive marker of overall health and serve as an indicator of high risk for exacerbation of HIV-related and comorbid conditions. Predicting hospitalization risk with a risk score can combine HIV-related clinical status, comorbid conditions, and social determinants of health (SDOH) into one measure, and can lead to interventions that prevent hospitalizations, alter clinical progression, and improve patient health. We do not yet adequately understand hospitalization causes or predictors of hospitalization among PWH. Therefore, in Aim 1, we will develop and implement a data-driven algorithm, incorporating a standardized adjudication protocol, for assigning validated hospitalization causes. We will use these validated causes to estimate trends and patterns in cause-specific hospitalization rates across calendar years and patient characteristics. In Aim 2, we will develop hospitalization risk scores for use in outpatient HIV care, using new innovative interpretable machine learning algorithms with integrated clinical, patient reported outcome and SDOH data. In Aim 3, we will assess feasibility, acceptability, and usefulness of using hospitalization risk scores as a clinical support tool in outpatient HIV clinical care via formative research. Stakeholders will be engaged through focus groups and semi-structured interviews. Stakeholder perspectives will be sought from patients, clinical providers, hospital and clinic administrators, epidemiologists, computer, data, and implementation scientists, and an ethicist. These aims have potential to improve the health of PWH across the lifespan, leveraging advances in epidemiology and data science and translating these to support HIV outpatient clinical care.
