. Early HIV diagnosis as well as linkage into and retention in HIV medical care for HIV+ individuals is important for patient survival and treatment. Missed opportunities for early HIV diagnosis continues even with recommended routine HIV testing. National and South Carolina (SC) estimates of retention in HIV medical care are slightly above fifty percent, indicating a gap in HIV treatment. With significant proportions of HIV+ individuals not receiving HIV medical care, improved outcomes of care and HIV prevention as part of national HIV/AIDs strategies are difficult to achieve. The purpose of this study is to use novel machine learning algorithms to further explore, identify, characterize, and explain predictors of missed opportunities for HIV medical care utilization among all living HIV+ individuals in SC. Profiles of HIV+ individuals based on their patterns of HIV medical care seeking behavior will be developed with concomitant identification of both gaps in HIV care and missed opportunities for reengagement into HIV care. Health utilization behavior for HIV+ individuals' pre-HIV diagnosis also will be studied to identify where missed opportunities for HIV testing occurs. Findings will be integrated with the ongoing effort of the SC Department of Health and Environmental Control (DHEC)'s Data-to-Care (DTC) program as well as the Ryan White Care Program. The public health value that HIV treatment brings includes improved survival outcomes of care among HIV+ individuals as well as reduced HIV transmission. These important components form part of the overall strategy for fighting and controlling the HIV epidemic in the United States and aligns closely with the strategic goals of reducing new HIV infections. Using state-level CD4 and Viral Load (VL) testing data available for all SC HIV+ individuals since 2004, the study will link inpatient and outpatient claims data sources, the state electronic HIV/AIDS reporting system, Area Health Resource Files, and data from the state corrections database to create a unique population based dataset spanning 10 years (2004-2013). Advanced Big Data analytical algorithms will be used to create person-level profile patterns of pre- and post- HIV diagnosis health utilization behaviors and for identifying best predictors of linkage and retention in HIV medical care. These algorithms will be useful in unearthing hidden features/predictors of HIV medical care utilization. A predictive model useful for predicting where HIV+ individuals who are not in care will access routine medical care (missed opportunities) also will be developed. Findings will provide fresh guidance for public health interventions targeting early HIV testing and linkage to and retention in HIV medical care for SC HIV-infected individuals.
. Treatment as a public health strategy of HIV prevention is important in fighting the national HIV epidemic. Early diagnosis, linkage to, and retention in HIV medical care are key public health strategies for HIV prevention. Missed opportunities for HIV diagnosis as well as inadequate HIV medical care utilization by HIV+ individuals in South Carolina negatively impacts HIV survival outcomes, while increasing population transmission risks. The analyses of population-based health utilization data for HIV+ individuals using powerful machine learning techniques such as deep learning algorithms will generate new factors useful for improving retention in HIV care. This valuable new information will be translated into ongoing efforts such as SC's Data-to Care and Ryan White care efforts through improved targeting of HIV medical care engagement and reengagement interventions.
|Zhang, Qingpeng; Zhong, Lu; Gao, Siyang et al. (2018) Optimizing HIV Interventions for Multiplex Social Networks via Partition-Based Random Search. IEEE Trans Cybern 48:3411-3419|