Using big data analytics to improve HIV medical care utilisation in South Carolina: A study protocol

Association Between Dynamic Viral Rebound and Longitudinal Measures of Viral Load/CD4 Counts Among People with HIV in South Carolina

Monitoring HIV viral rebound (VR) is crucial, as it indicates an increased risk of infection, transmission, disease progression, and drug resistance. This study aims to identify the association between dynamic VR and historical viral load (VL)/CD4 count measures. Fifteen-year South Carolina population-based electronic health record data were used for the study. VR was defined as the return of detectable levels of VL (>200 copies/mL) after stable viral suppression (VS) (two consecutive VS, i.e., VL ≤200 copies/mL). A generalized linear mixed model was used to evaluate the association between dynamic VR and historical time-dependent predictors, such as nadir CD4 count and comorbidities, within a year prior to each VR. Subgroup analysis for men who have sex with men (MSM) was also conducted. Among 8,185 people with HIV (PWH), 1,173 (14.3%) had a history of VR. Lower nadir CD4 count (≥500 vs. <200 cells/µL; adjusted odds ratio [aOR]: 0.51, 95% confidence interval [CI]: [0.43, 0.60]), younger age (>60 years old vs. 18-30 years old; aOR: 0.43, 95% CI: [0.29, 0.63]), and being Black (Black vs. White; aOR: 1.58, 95% CI: [1.34, 1.85]) were associated with a higher risk of VR, while MSM (MSM vs. heterosexual; aOR: 0.81, 95% CI: [0.67, 0.96]) were associated with decreased VR risk. The rate of VR among PWH in South Carolina is significant. Within-1-year VL/CD4 test is critical for identifying PWH at risk for VR. Tailored interventions are needed for PWH at risk for VR to achieve sustained suppression and better health outcomes.

The future of HIV diagnostics: an exemplar in infectious diseases

Over the past 40 years, diagnostics have become the backbone of HIV prevention, treatment, and retention in care, and are central to the achievement of UNAIDS 95-95-95 targets. Over the next decade, the global HIV response will face difficult challenges. In addition to sustaining gains achieved in prevention and treatment, substantial gaps in care need to be addressed for underserved populations. Diagnostics will play an important role in control and prevention of HIV infection through novel technologies, digital solutions, and integrated service delivery innovations. The integration of diagnostics with digital health, machine learning, and generative artificial intelligence provides opportunities for more effective individual and public health disease control. These diagnostics and other futuristic innovations such as wearable technologies, omics, metaverse-based solutions, and quantum diagnostics could enable the achievement of the UNAIDS 95-95-95 targets; however, their use will face barriers related to health-care system financing, infrastructure, technological readiness and skills, and long-term sustainability. This Review highlights diagnostic strategies and innovations that could catalyse a new era in the management of the HIV pandemic.

Association between substance use disorders and sustained viral suppression: a longitudinal analysis among people with HIV in South Carolina

OBJECTIVES

Substance use disorders (SUDs) are a significant public health concern across the United States and may pose a risk to achieving sustained viral suppression (SVS) in people with HIV (PWH). This study aims to examine the association between SUDs and SVS among PWH.

DESIGN

Using electronic health records from the South Carolina Department of Health, we conducted a retrospective study of adults with HIV who were diagnosed between January 2006 and December 2019.

METHODS

The impact of SUDs on SVS was assessed using generalized linear mixed model. Potential confounders included age, sex, chronic diseases history, etc. Stepwise selection was performed to decide the confounders included in the final model, and the optimal correlation structure was determined by Akaike information criterion.

RESULTS

Of the 9412 eligible participants, 7481 (79.48%) had reached SVS status during their follow-up periods. SUDs related to alcohol [adjusted odds ratio (AOR) = 1.70, 95% confidence interval (CI): 1.46-1.98], cannabis (AOR = 1.62, 95% CI: 1.35-1.95), cocaine (AOR = 1.95, 95% CI: 1.60-2.37), opioid (AOR = 1.91, 95% CI: 1.13-3.23), and tobacco (AOR = 1.80, 95% CI: 1.69-1.92) were negatively associated with SVS. Individuals with chronic conditions such as cardiovascular disease (AOR = 0.31, 95% CI: 0.29-0.33), diabetes (AOR = 0.49, 95% CI: 0.41-0.59), and cancer (AOR = 0.47, 95% CI: 0.38-0.58) showed a higher likelihood of maintaining SVS.

CONCLUSION

This large cohort study of PWH with extended follow-up highlights the negative impact of SUDs on maintaining SVS. Long-term strategies for reducing substance use could support SVS in PWH.

Using Machine Learning Techniques to Predict Viral Suppression Among People With HIV

BACKGROUND

This study aims to develop and examine the performance of machine learning (ML) algorithms in predicting viral suppression among statewide people living with HIV (PWH) in South Carolina.

METHODS

Extracted through the electronic reporting system in South Carolina, the study population was adult PWH who were diagnosed between 2005 and 2021. Viral suppression was defined as viral load <200 copies/mL. The predictors, including sociodemographics, a historical information of viral load indicators (eg, viral rebound), comorbidities, health care utilization, and annual county-level factors (eg, social vulnerability), were measured in each 4-month windows. Using historic information in different lag time windows (1-, 3- or 5-lagged time windows with each 4-month window as a unit), both traditional and ML approaches (eg, Long Short-Term Memory Network) were applied to predict viral suppression. Comparisons of prediction performance between different models were assessed by area under curve (AUC), recall, precision, F1 score, and Youden index.

RESULTS

ML approaches outperformed the generalized linear mixed model. In all the 3 lagged analysis of a total of 15,580 PWH, the Long Short-Term Memory Network (Lag 1: AUC = 0.858; Lag 3: AUC = 0.877; Lag 5: AUC = 0.881) algorithm outperformed all the other methods in terms of AUC performance for predicting viral suppression. The top-ranking predictors that were common in different models included historical information of viral suppression, viral rebound, and viral blips in the Lag-1 time window. Inclusion of county-level variables did not improve the model prediction accuracy.

CONCLUSIONS

Supervised ML algorithms may offer better performance for risk prediction of viral suppression than traditional statistical methods.

AI applications in HIV research: advances and future directions

With the increasing application of artificial intelligence (AI) in medical research, studies on the human immunodeficiency virus type 1(HIV-1) and acquired immunodeficiency syndrome (AIDS) have become more in-depth. Integrating AI with technologies like single-cell sequencing enables precise biomarker identification and improved therapeutic targeting. This review aims to explore the advancements in AI technologies and their applications across various facets of HIV research, including viral mechanisms, diagnostic innovations, therapeutic strategies, and prevention efforts. Despite challenges like data limitations and model interpretability, AI holds significant potential in advancing HIV-1 management and contributing to global health goals.

Longitudinal Viral Load Clustering for People With HIV Using Functional Principal Component Analysis

Background: Longitudinal measures of viral load (VL) are critical in monitoring the HIV status. While multiple lab indicators exist for monitoring measures of VL, research on clustering historical/longitudinal VL measures is limited. Analyzing longitudinal VL patterns rather than aggregated measures offers deeper insights into HIV status. This study uses functional data clustering to classify longitudinal VL patterns and characterize each cluster by demographics, comorbidities, social behaviors, and CD4 count. Methods: Adult PWH diagnosed from 2005 to 2015 in South Carolina with a 5-year minimum follow-up were included. We compared functional principal component analysis (FPCA), K-means, hierarchical clustering, and Gaussian mixture models for classification and found FPCA yielded the best results. ANOVA was used to compare VL characteristics, demographics, comorbidities, substance uses, and longitudinal CD4 count across clusters. Results: Results obtained from FPCA could best distinguish the characteristics and patterns into four clusters. A total of 5916 PWH were grouped into long-term VS group (Cluster 1, 17.3%), short-term VS group (Cluster 2, 29.8%), suboptimal VS group (Cluster 3, 28.3%), and viral failure group (Cluster 4, 24.9%). In the long-term VS group with an average of 11-year follow-up, PWH displayed sustained VS (95.3%) and lower mean CD4 count (95.3%) than other clusters. The short-term VS group had shorter follow-up (6 years), more comorbidities (31.4%), and lower percentage of time with low CD4 count (79.9%). In suboptimal VS group, PWH were mostly under 30 years old (44.8%) and Black (77.2%), with relatively lower mean VL (92.9%) and lower VR history (18.4%). In the viral failure group, PWH had higher mean VL (40.6%) and lower mean CD4 count (34.7%). Discussion: The findings highlight the impact of continuous clustering in understanding the distinct viral profiles of PWH and emphasize the importance of tailored treatment and insights to target interventions for all PWH.

Longitudinal assessments of viral rebound among people with HIV in South Carolina: a population-based cohort study

Routinely monitoring viral rebound (VR) is important in the life course of people with HIV (PWH). This study examined risk factors for time to the first VR, the number of VRs and their association with VR history in men who have sex with men (MSM). It includes 8176 adult PWH diagnosed from January 2005 to December 2018, followed until July 2021. We used the Cox model for time to the first VR, the Poisson model for a number of VRs, and logistic regression for VR history in MSM. Younger individuals (50-59 years vs 18-29 years, aHR: 0.43, 95% CI: [0.34, 0.55]) were more likely to experience VR. Black individuals (Black vs White, IRR: 1.61, 95% CI [1.38, 1.88]) had more VR, while MSM (MSM vs Heterosexual, IRR: 0.68, 95% CI: [0.57, 0.81]) was negatively associated with number of VsR. Furthermore, individuals engaging illicit drug use (IDU) (aOR: 1.50, 95% CI: [1.03, 2.17]) were more likely to experience VR in the MSM subgroup. This study highlighted the alarming risk factors related to VR among PWH. Tailored intervention should also be deployed for young, Black MSM patients with substance use for more effective and targeted public health strategies concerning VR.

Use of machine learning approaches to predict transition of retention in care among people living with HIV in South Carolina: a real-world data study

Maintaining retention in care (RIC) for people living with HIV (PLWH) helps achieve viral suppression and reduce onward transmission. This study aims to identify the best machine learning model that predicts the RIC transition over time. Extracting from the enhanced HIV/AIDS reporting system, this study included 9765 PLWH from 2005 to 2020 in South Carolina. Transition of RIC was defined as the change of RIC status in each two-year time window. We applied seven classifiers, such as Random Forest, Support Vector Machine, eXtreme Gradient Boosting and Long-short-term memory, for each lagged response to predict the subsequent year's RIC transition. Classification performance was assessed using balanced prediction accuracy, the area under the curve (AUC), recall, precision and F1 scores. The proportion of the four categories of RIC transition was 13.59%, 29.78%, 9.06% and 47.57%, respectively. Support Vector Machine was the best approach for every lag model based on both the F1 score (0.713, 0.717 and 0.719) and AUC (0.920, 0.925 and 0.928). The findings could facilitate the risk augment of PLWH who are prone to follow-up so that clinicians and policymakers could come up with more specific strategies and relocate resources for intervention to keep them sustained in HIV care.

The Impacts of HIV-Related Service Interruptions During the COVID-19 Pandemic: Protocol of a Mixed Methodology Longitudinal Study

The global COVID-19 pandemic has imposed unprecedented pressure on health systems and has interrupted public health efforts for other major health conditions, including HIV. It is critical to comprehensively understand how the pandemic has affected the delivery and utilization of HIV-related services and what are the effective strategies that may mitigate the negative impacts of COVID-19 and resultant interruptions. The current study thus aims to comprehensively investigate HIV service interruptions during the pandemic following a socioecological model, to assess their impacts on various outcomes of the HIV prevention and treatment cascade and to identify resilience resources for buffering impacts of interruptions on HIV treatment cascade outcomes. We will assess HIV service interruptions in South Carolina (SC) since 2020 using operational report data from Ryan White HIV clinics and HIV service utilization data (including telehealth use) based on statewide electronic health records (EHR) and cellphone-based place visitation data. We will further explore how HIV service interruptions affect HIV prevention and treatment cascade outcomes at appropriate geospatial units based on the integration of multi-type, multi-source datasets (e.g., EHR, geospatial data). Finally, we will identify institutional-, community-, and structural-level factors (e.g., resilience resources) that may mitigate the adverse impacts of HIV service interruptions based on the triangulation of quantitative (i.e., EHR data, geospatial data, online survey data) and qualitative (i.e., in-depth interviews with clinic leaders, healthcare providers, people living with HIV, and HIV clinic operational reports) data regarding health system infrastructure, social capital, and organizational preparedness. Our proposed research can lead to a better understanding of complicated HIV service interruptions in SC and resilience factors that can mitigate the negative effects of such interruptions on various HIV treatment cascade outcomes. The multilevel resilience resources identified through data triangulation will assist SC health departments and communities in developing strategic plans in response to this evolving pandemic and other future public health emergencies (e.g., monkeypox, disasters caused by climate change). The research findings can also inform public health policymaking and the practices of other Deep South states with similar sociocultural contexts in developing resilient healthcare systems and communities and advancing epidemic preparedness.

Functional Multivariable Logistic Regression With an Application to HIV Viral Suppression Prediction

Motivated by improving the prediction of the human immunodeficiency virus (HIV) suppression status using electronic health records (EHR) data, we propose a functional multivariable logistic regression model, which accounts for the longitudinal binary process and continuous process simultaneously. Specifically, the longitudinal measurements for either binary or continuous variables are modeled by functional principal components analysis, and their corresponding functional principal component scores are used to build a logistic regression model for prediction. The longitudinal binary data are linked to underlying Gaussian processes. The estimation is done using penalized spline for the longitudinal continuous and binary data. Group-lasso is used to select longitudinal processes, and the multivariate functional principal components analysis is proposed to revise functional principal component scores with the correlation. The method is evaluated via comprehensive simulation studies and then applied to predict viral suppression using EHR data for people living with HIV in South Carolina.

The dynamic risk factors of cardiovascular disease among people living with HIV: a real-world data study

BACKGROUND

This study aims to investigate the incidence and dynamic risk factors for cardiovascular diseases (CVD) among people living with HIV (PLWH).

METHODS

In this population-based statewide cohort study, we utilized integrated electronic health records data to identify adult (age ≥ 18) who were diagnosed with HIV between 2006 and 2019 and were CVD event-free at the HIV diagnosis in South Carolina. The associations of HIV-related factors and traditional risk factors with the CVD incidence were investigated during the overall study period, and by different follow-up periods (i.e., 0-5yrs, 6-10yrs 11-15yrs) using multivariable logistic regression models.

RESULTS

Among 9,082 eligible participants, the incidence of CVD was 18.64 cases per 1000 person-years. Overall, conventional risk factors, such as tobacco use, hypertension, obesity, chronic kidney disease (CKD), were persistently associated with the outcome across all three groups. While HIV-related factors, such as recent CD4 count (e.g., > 350 vs. <200 cells/mm3: adjusted odds ratio [aOR] range: 0.18-0.25), and percent of years in retention (e.g., 31-75% vs. 0-30%: aOR range: 0.24-0.57) were associated with lower odds of CVD incidence regardless of different follow up periods. The impact of the percent of days with viral suppression gradually diminished as the follow-up period increased.

CONCLUSIONS

Maintaining an optimal viral suppression might prevent CVD incidence in the short term, whereas restoring immune recovery may be beneficial for reducing CVD risk regardless of the duration of HIV diagnosis. Our findings suggest the necessity of conducting more targeted interventions during different periods of HIV infection.

Suicide Ideation and Attempt Among People With HIV: A Statewide Population-level Cohort Analysis Between 2005 and 2020

BACKGROUND

Risk factors for suicidality among people with HIV (PWH) may evolve over their disease course, particularly as they develop comorbidities such as mental health disorders over time.

SETTING

This study compared the leading risk factors of suicide ideation/attempt among PWH in South Carolina across different combination antiretroviral therapy (cART) eras.

METHODS

A statewide cohort of PWH who were diagnosed between 2005 and 2016, with a follow-up record until 2020, was involved in the study. A Cox proportional hazards model was used to examine the association of suicide ideation/attempt and predictors, including demographics, HIV-related characteristics, and mental health conditions.

RESULTS

Among 8567 PWH, the incidence of suicide ideation/attempt increased from 537.7 per 100,000 person-years [95% confidence interval (CI): 460.2 to 615.1] in the early cART cohort (2005-2008) to 782.5 (95% CI: 697.6 to 867.4) in the late cART cohort (2009-2016). Leading risk factors of suicide ideation/attempt changed across the cART cohort. In the early cART cohort, PWH with suicide ideation/attempt were more likely to be White and diagnosed with bipolar disorder ( P' s < 0.05). In the late cART cohort, suicide ideation/attempt was positively associated with transmission through injection drug use, anxiety, posttraumatic stress disorder, schizophrenia, and personality disorder ( P' s < 0.05).

CONCLUSIONS

Mental health conditions have emerged as more prominent risk factors for suicide ideation/attempt in the late cART cohort. Enhanced access to psychiatric care could facilitate the early identification of mental health conditions, enabling timely counseling or psychosocial interventions that may mitigate mental health issues and, consequently, reduce the likelihood of suicide ideation/attempts among PWH.

Retrospective Review of Complications and Revision Rates Between Isolated Talonavicular vs Talonavicular and Subtalar (Double) Arthrodesis vs Triple Arthrodesis

Background

Hindfoot fusion procedures are common for the treatment of end-stage arthritis or deformity. Surgical treatments for these conditions include talonavicular joint (single) arthrodesis, talonavicular and subtalar (double) arthrodesis, or talonavicular, subtalar, and calcaneocuboid (triple) arthrodesis. This study evaluated the complication rate, revision surgery rate, and hardware removal rate for those treated with either single, double, or triple arthrodesis.

Methods

A retrospective review was conducted for patients who underwent single (Current Procedural Terminology [CPT] code 28740), double (CPT 28725 and 28740), or triple (CPT 28715) arthrodesis to treat hindfoot arthritis/deformity (International Classification of Diseases, Ninth Revision [ICD-9] code: 734, International Classification of Diseases, Tenth Revision [ICD-10] codes: M76821, M76822, and M76829) from 2005 to 2022 using the South Carolina Revenue and Fiscal Affairs databank. Data collected included demographics, comorbidities, procedure data, and postoperative outcomes within 1 year of principal surgery. Student t test, chi-squared test, and multivariable logistic regression analysis were utilized during data analysis.

Results

A total of 433 patients were identified, with 248 undergoing single arthrodesis, 67 undergoing double arthrodesis, and 118 undergoing triple arthrodesis. There was no significant difference between single, double, and triple arthrodesis in the rate of complications, hardware removals, revision surgeries, or 30-day readmission when controlling for confounding variables. However, a decrease in Charlson Comorbidity Index (CCI) was found to be predictive of an increase in the revision surgery rate (OR = 0.46, 95% CI 0.22-0.85, P = .02).

Conclusion

We found no difference in the rate of complications, hardware removals, or revision surgeries in those undergoing single, double, or triple arthrodesis. Surprisingly we found that a lower Charlson Comorbidity Index, indicating a healthier patient had a significant relationship with a higher rate of revision surgery. Further study including radiographic indications for surgery or the impact of overall health status on revision surgery rates may further elucidate the other components of this relationship.

Level of Evidence

Level III, cohort study.

Patterns and predictors of racial/ethnic disparities in HIV care continuum in the Southern USA: protocol for a population-based cohort study

INTRODUCTION

Health disparities exist at every step of the HIV care continuum (HCC) among racial/ethnic minority population. Such racial/ethnic disparities may have significantly delayed the progress in HCC in the Southern US states that are strongly represented among geographic focus areas in the 2019 federal initiative titled 'Ending the HIV Epidemic: A Plan for America'. However, limited efforts have been made to quantify the long-term spatiotemporal variations of HCC disparities and their contributing factors over time, particularly in the context of COVID-19 pandemic. This project aims to identify the spatiotemporal patterns of racial disparities of each HCC outcome and then determine the contribution of contextual features for temporal change of disparities in HCC.

METHODS AND ANALYSIS

This cohort study will use statewide HIV cohort data in South Carolina, including all people living with HIV (PLWH) who were diagnosed with HIV in 2005-2020. The healthcare encounter data will be extracted from longitudinal EHR from six state agencies and then linked to aggregated county-level community and social structural-level data (eg, structural racism, COVID-19 pandemic) from multiple publicly available data sources. The South Carolina Revenue of Fiscal and Affairs will serve as the honest broker to link the patient-level and county-level information. We will first quantify the HCC-related disparities by creating a county-level racial/ethnic disparity index (RDI) for each key HCC outcomes (eg, HIV testing, timely diagnosis), examine the temporal patterns of each RDI over time and then using geographical weighted lasso model examine which contextual factors have significant impacts on the change of county-level RDI from 2005 to 2020.

ETHICS AND DISSEMINATION

The study was approved by the Institutional Review Board at the University of South Carolina (Pro00121718) as a Non-Human Subject study. The study's findings will be published in peer-reviewed journals and disseminated at national and international conferences and through social media.

Association of CD4 + cell count and HIV viral load with risk of non-AIDS-defining cancers

OBJECTIVES

HIV-induced immunodeficiency contributes to an increased risk of non-AIDS-defining cancers (NADC). This study aims to identify the most predictive viral load (VL) or CD4 + measures of NADC risk among people with HIV (PWH).

DESIGN

Extracted from South Carolina electronic HIV reporting system, we studied adult PWH who were cancer-free at baseline and had at least 6 months of follow-up since HIV diagnosis between January 2005 and December 2020.

METHODS

Using multiple proportional hazards models, risk of NADC was investigated in relation to 12 measures of VL and CD4 + cell count at three different time intervals before NADC diagnosis. The best VL/CD4 + predictor(s) and final model were determined using Akaike's information criterion.

RESULTS

Among 10 413 eligible PWH, 449 (4.31%) developed at least one type of NADC. After adjusting for potential confounders, the best predictors of NADC were the proportion of days with viral suppression (hazard ratio [HR]: 0.47 (>25% and ≤50% vs. 0), 95% confidence interval [CI]: [0.28, 0.79]) and proportion of days with low CD4 + cell count (AIC = 7201.35) (HR: 12.28 (>75% vs. = 0), 95% CI: [9.29, 16.23]).

CONCLUSIONS

VL and CD4 + measures are strongly associated with risk of NADC. In analyses examining three time windows, proportion of days with low CD4 + cell count was the best CD4 + predictor for each time window. However, the best VL predictor varied across time windows. Thus, using the best combination of VL and CD4 + measures for a specific time window should be considered when predicting NADC risk.

The incidence and dynamic risk factors of chronic kidney disease among people with HIV

OBJECTIVES

We investigate the incidence of chronic kidney disease (CKD) among people with HIV (PWH) and the dynamic risk factors associated with CKD incidence.

DESIGN

A population-based cohort study of PWH in South Carolina.

METHODS

Adults (age ≥18 years) PWH diagnosed between 2006 and 2019 who were CKD-free at baseline were included. The associations of HIV-related risk factors and conventional risk factors with the incidence of CKD were investigated during the overall study period and by different follow-up periods (i.e. 5, 10, and 15 years) by multivariate logistic regression.

RESULTS

Among 9514 PWH, the incidence of CKD was 12.39 per 1000 person-years. The overall model indicated that conventional risk factors, such as hypertension, dyslipidemia, cardiovascular disease, and diabetes, were significantly associated with a higher risk of developing CKD. HIV-related characteristics, such as high percentage of days with viral suppression, recent CD4 + cell count, and percentage of retention in care, were associated with a lower risk of CKD compared with their counterparts. In the subgroup analysis, the results were similar for the 5-year and 6-10 years follow-up groups. Among patients who did not develop CKD by the 10th year, the risk factors for developing CKD within 11-15 years were dyslipidemia, diabetes, low recent CD4 + cell count, and short duration of retention in care while other predictors vanished.

CONCLUSION

Diabetes, CD4 + cell count, and retention in care were persistently associated with CKD despite of follow-up duration. Closely monitoring diabetes and improving CD4 + cell count and retention in care are important to lower the risk of CKD in PWH.

County-level variations in linkage to care among people newly diagnosed with HIV in South Carolina: A longitudinal analysis from 2010 to 2018

BACKGROUND

Timely linkage to care (LTC) is key in the HIV care continuum, as it enables people newly diagnosed with HIV (PNWH) to benefit from HIV treatment at the earliest stage. Previous studies have found LTC disparities by individual factors, but data are limited beyond the individual level, especially at the county level. This study examined the temporal and geographic variations of county-level LTC status across 46 counties in South Carolina (SC) from 2010 to 2018 and the association of county-level characteristics with LTC status.

METHODS

All adults newly diagnosed with HIV from 2010 to 2018 in SC were included in this study. County-level LTC status was defined as 1 = "high LTC (≥ yearly national LTC percentage)" and 0 = "low LTC (< yearly national LTC percentage)". A generalized estimating equation model with stepwise selection was employed to examine the relationship between 29 county-level characteristics and LTC status.

RESULTS

The number of counties with high LTC in SC decreased from 34 to 21 from 2010 to 2018. In the generalized estimating equation model, six out of 29 factors were significantly associated with LTC status. Counties with a higher percentage of males (OR = 0.07, 95%CI: 0.02~0.29) and persons with at least four years of college (OR = 0.07, 95%CI: 0.02~0.34) were less likely to have high LTC. However, counties with more mental health centers per PNWH (OR = 45.09, 95%CI: 6.81~298.55) were more likely to have high LTC.

CONCLUSIONS

Factors associated with demographic characteristics and healthcare resources contributed to the variations of LTC status at the county level. Interventions targeting increasing the accessibility to mental health facilities could help improve LTC.

Prevalence and trend of AIDS-defining cancers and non-AIDS-defining cancers and their association with antiretroviral therapy among people living with HIV in South Carolina: a population-based cohort study

ABSTRACTMonitoring cancer trends and risk is critical as cancer remains a growing problem in persons living with HIV (PLWH). Recent population-based data are limited regarding the cancer trends among PLWH. Our study examined the prevalence and trends in the rate of AIDS-defining cancers (ADC) and non-AIDS-defining cancers (NADC) and their risk factors in PLWH in South Carolina. Utilizing linked population-based HIV data (2005-2020), time-dependent proportional hazards model was used to identify associated risk predictors of developing cancer in PLWH. Among 11,238 PLWH, 250 individuals developed ADC and 454 developed NADC. The median time from HIV diagnosis to cancer diagnosis was 1.9 years for ADC and 3.8 years for NADC. Individuals who developed ADC or NADC were more likely to be older, male, use substances, have hepatitis infection, hypothyroidism, hypertension, and renal disease. Individuals with viral load >100,000 copies/ml were more likely to develop ADC while those with CD4 count >350 cells/mm3 were less likely to develop ADC or NADC. Our findings suggest that long-term viral suppression may contribute to risk reduction for cancer in PLWH. Early HIV diagnosis along with viral load suppression should be a part of ongoing cancer prevention efforts.

The impact of COVID-19 pandemic on the dynamic HIV care engagement among people with HIV: real-world evidence

INTRODUCTION

Existing studies examining the impact of the COVID-19 pandemic on engagement in HIV care often capture cross-sectional status, while lacking longitudinal evaluations. This study examined the impact of the pandemic on the longitudinal dynamic change of retention in care and viral suppression status.

METHODS

The electronic health record (EHR) data of this population-level cohort study were retrieved from the statewide electronic HIV/AIDS reporting system in South Carolina. The study population was people with HIV (PWH) who had at least one year's symmetric follow-up observation record before and after the pandemic. Multivariable generalized linear mixed regression models were employed to analyze the impact of the pandemic on these outcomes, adjusting for socio-demographic characteristics and preexisting comorbidities.

RESULTS

In the adjusted models, PWH had a lower likelihood of retention in care (adjusted odds ratio [aOR]: 0.806, 95% confidence interval [CI]: 0.769, 0.844) and a higher probability of virological failure (aOR: 1.240, 95% CI: 1.169, 1.316) during the peri-pandemic period than pre-pandemic period. Results from interaction effect analysis from each cohort revealed that the negative effect of the pandemic on retention in care was more severe among PWH with high comorbidity burden than those without any comorbidity; meanwhile, a more striking virological failure was observed among PWH who reside in urban areas than in rural areas.

CONCLUSION

The COVID-19 pandemic has a negative impact on retention in care and viral suppression among PWH in South Carolina, particularly for individuals with comorbidities and residing in urban areas.

Moderation effect of community health on the relationship between racial/ethnic residential segregation and HIV viral suppression in South Carolina: A county-level longitudinal study from 2013 to 2018

Background

Viral suppression is the ultimate goal of the HIV treatment cascade and a primary endpoint of antiretroviral therapy. Empirical evidence found racial/ethnic disparities in viral suppression among people living with HIV (PWH), but the evidence of the relationship between racial/ethnic residential segregation and place-based viral suppression is scarce. Further exploring potential structural moderators in this relationship has substantial implications for healthcare policymaking and resource allocation. The current study aimed to investigate the spatial-temporal disparities in the HIV viral suppression rate across 46 counties in South Carolina from 2013 to 2018. We also examined the impact of racial/ethnic residential segregation and the moderation effect of community health, one measurement of community engagement and volunteerism.

Methods

The proportion of PWH who achieved viral suppression for each county and calendar year was calculated using de-identified electronic medical records. The isolation index was calculated and used to measure racial/ethnic residential segregation. The community health index and other county-level factors were directly extracted from multiple publicly available datasets. We used geospatial mapping to explore the spatial-temporal variations of HIV viral suppression rates. Hierarchical quasi-binominal regression models were used to examine the impacts of racial/ethnic residential segregation on county-level viral suppression rate by the extent of community health.

Results

From 2013 to 2018, the average viral suppression rate across 46 counties in SC increased from 64.3% to 65.4%. Regression results revealed that counties with high racial/ethnic residential segregation were more likely to have a low viral suppression rate (β = -0.56, 95% CI: -0.75 to -0.37). In counties with high levels of community health, the impact of racial/ethnic residential segregation on viral suppression rate decreased as compared with those with low levels of community health (β = 5.50, 95% CI: 0.95-10.05).

Conclusions

Racial/ethnic residential segregation acts as a structural barrier to placed-based viral suppression rates and compromises the goal of the HIV treatment cascade. Concentrated and sustained county-level interventions aiming to improve community health can be practical approaches to promote health equity in HIV treatment and care.

Multicenter Development and Validation of a Model for Predicting Retention in Care Among People with HIV

Predictive analytics can be used to identify people with HIV currently retained in care who are at risk for future disengagement from care, allowing for prioritization of retention interventions. We utilized machine learning methods to develop predictive models of retention in care, defined as no more than a 12 month gap between HIV care appointments in the Center for AIDS Research Network of Integrated Clinical Systems (CNICS) cohort. Data were split longitudinally into derivation and validation cohorts. We created logistic regression (LR), random forest (RF), and gradient boosted machine (XGB) models within a discrete-time survival analysis framework and compared their performance to a baseline model that included only demographics, viral suppression, and retention history. 21,267 Patients with 507,687 visits from 2007 to 2018 were included. The LR model outperformed the baseline model (AUC 0.68 [0.67-0.70] vs. 0.60 [0.59-0.62], P < 0.001). RF and XGB models had similar performance to the LR model. Top features in the LR model included retention history, age, and viral suppression.

Oncological big data platforms for promoting digital competencies and professionalism in Chinese medical students: a cross-sectional study

OBJECTIVES

Advancements in big data technology are reshaping the healthcare system in China. This study aims to explore the role of medical big data in promoting digital competencies and professionalism among Chinese medical students.

DESIGN, SETTING AND PARTICIPANTS

This study was conducted among 274 medical students who attended a workshop on medical big data conducted on 8 July 2021 in Tongji Hospital. The workshop was based on the first nationwide multifunction gynecologic oncology medical big data platform in China, at the National Union of Real-World Gynecologic Oncology Research & Patient Management Platform (NUWA platform).

OUTCOME MEASURES

Data on knowledge, attitudes towards big data technology and professionalism were collected before and after the workshop. We have measured the four skill categories: doctor‒patient relationship skills, reflective skills, time management and interprofessional relationship skills using the Professionalism Mini-Evaluation Exercise (P-MEX) as a reflection for professionalism.

RESULTS

A total of 274 students participated in this workshop and completed all the surveys. Before the workshop, only 27% of them knew the detailed content of medical big data platforms, and 64% knew the potential application of medical big data. The majority of the students believed that big data technology is practical in their clinical practice (77%), medical education (85%) and scientific research (82%). Over 80% of the participants showed positive attitudes toward big data platforms. They also exhibited sufficient professionalism before the workshop. Meanwhile, the workshop significantly promoted students' knowledge of medical big data (p<0.05), and led to more positive attitudes towards big data platforms and higher levels of professionalism.

CONCLUSIONS

Chinese medical students have primitive acquaintance and positive attitudes toward big data technology. The NUWA platform-based workshop may potentially promote their understanding of big data and enhance professionalism, according to the self-measured P-MEX scale.

Multilevel determinants of racial/ethnic disparities in severe maternal morbidity and mortality in the context of the COVID-19 pandemic in the USA: protocol for a concurrent triangulation, mixed-methods study

INTRODUCTION

The COVID-19 pandemic has affected communities of colour the hardest. Non-Hispanic black and Hispanic pregnant women appear to have disproportionate SARS-CoV-2 infection and death rates.

METHODS AND ANALYSIS

We will use the socioecological framework and employ a concurrent triangulation, mixed-methods study design to achieve three specific aims: (1) examine the impacts of the COVID-19 pandemic on racial/ethnic disparities in severe maternal morbidity and mortality (SMMM); (2) explore how social contexts (eg, racial/ethnic residential segregation) have contributed to the widening of racial/ethnic disparities in SMMM during the pandemic and identify distinct mediating pathways through maternity care and mental health; and (3) determine the role of social contextual factors on racial/ethnic disparities in pregnancy-related morbidities using machine learning algorithms. We will leverage an existing South Carolina COVID-19 Cohort by creating a pregnancy cohort that links COVID-19 testing data, electronic health records (EHRs), vital records data, healthcare utilisation data and billing data for all births in South Carolina (SC) between 2018 and 2021 (>200 000 births). We will also conduct similar analyses using EHR data from the National COVID-19 Cohort Collaborative including >270 000 women who had a childbirth between 2018 and 2021 in the USA. We will use a convergent parallel design which includes a quantitative analysis of data from the 2018-2021 SC Pregnancy Risk Assessment and Monitoring System (unweighted n>2000) and in-depth interviews of 40 postpartum women and 10 maternal care providers to identify distinct mediating pathways.

ETHICS AND DISSEMINATION

The study was approved by institutional review boards at the University of SC (Pro00115169) and the SC Department of Health and Environmental Control (DHEC IRB.21-030). Informed consent will be provided by the participants in the in-depth interviews. Study findings will be disseminated with key stakeholders including patients, presented at academic conferences and published in peer-reviewed journals.

Studying patterns and predictors of HIV viral suppression using A Big Data approach: a research protocol

Background

Given the importance of viral suppression in ending the HIV epidemic in the US and elsewhere, an optimal predictive model of viral status can help clinicians identify those at risk of poor viral control and inform clinical improvements in HIV treatment and care. With an increasing availability of electronic health record (EHR) data and social environmental information, there is a unique opportunity to improve our understanding of the dynamic pattern of viral suppression. Using a statewide cohort of people living with HIV (PLWH) in South Carolina (SC), the overall goal of the proposed research is to examine the dynamic patterns of viral suppression, develop optimal predictive models of various viral suppression indicators, and translate the models to a beta version of service-ready tools for clinical decision support.

Methods

The PLWH cohort will be identified through the SC Enhanced HIV/AIDS Reporting System (eHARS). The SC Office of Revenue and Fiscal Affairs (RFA) will extract longitudinal EHR clinical data of all PLWH in SC from multiple health systems, obtain data from other state agencies, and link the patient-level data with county-level data from multiple publicly available data sources. Using the deidentified data, the proposed study will consist of three operational phases: Phase 1: “Pattern Analysis” to identify the longitudinal dynamics of viral suppression using multiple viral load indicators; Phase 2: “Model Development” to determine the critical predictors of multiple viral load indicators through artificial intelligence (AI)-based modeling accounting for multilevel factors; and Phase 3: “Translational Research” to develop a multifactorial clinical decision system based on a risk prediction model to assist with the identification of the risk of viral failure or viral rebound when patients present at clinical visits.

Discussion

With both extensive data integration and data analytics, the proposed research will: (1) improve the understanding of the complex inter-related effects of longitudinal trajectories of HIV viral suppressions and HIV treatment history while taking into consideration multilevel factors; and (2) develop empirical public health approaches to achieve ending the HIV epidemic through translating the risk prediction model to a multifactorial decision system that enables the feasibility of AI-assisted clinical decisions.

Leveraging National Claims and Hospital Big Data: Cohort Study on a Statin-Drug Interaction Use Case

Background

Linking different sources of medical data is a promising approach to analyze care trajectories. The aim of the INSHARE (Integrating and Sharing Health Big Data for Research) project was to provide the blueprint for a technological platform that facilitates integration, sharing, and reuse of data from 2 sources: the clinical data warehouse (CDW) of the Rennes academic hospital, called eHOP (entrepôt Hôpital), and a data set extracted from the French national claim data warehouse (Système National des Données de Santé [SNDS]).

Objective

This study aims to demonstrate how the INSHARE platform can support big data analytic tasks in the health field using a pharmacovigilance use case based on statin consumption and statin-drug interactions.

Methods

A Spark distributed cluster-computing framework was used for the record linkage procedure and all analyses. A semideterministic record linkage method based on the common variables between the chosen data sources was developed to identify all patients discharged after at least one hospital stay at the Rennes academic hospital between 2015 and 2017. The use-case study focused on a cohort of patients treated with statins prescribed by their general practitioner or during their hospital stay.

Results

The whole process (record linkage procedure and use-case analyses) required 88 minutes. Of the 161,532 and 164,316 patients from the SNDS and eHOP CDW data sets, respectively, 159,495 patients were successfully linked (98.74% and 97.07% of patients from SNDS and eHOP CDW, respectively). Of the 16,806 patients with at least one statin delivery, 8293 patients started the consumption before and continued during the hospital stay, 6382 patients stopped statin consumption at hospital admission, and 2131 patients initiated statins in hospital. Statin-drug interactions occurred more frequently during hospitalization than in the community (3800/10,424, 36.45% and 3253/14,675, 22.17%, respectively; P<.001). Only 121 patients had the most severe level of statin-drug interaction. Hospital stay burden (length of stay and in-hospital mortality) was more severe in patients with statin-drug interactions during hospitalization.

Conclusions

This study demonstrates the added value of combining and reusing clinical and claim data to provide large-scale measures of drug-drug interaction prevalence and care pathways outside hospitals. It builds a path to move the current health care system toward a Learning Health System using knowledge generated from research on real-world health data.

Emergence and evolution of big data science in HIV research: Bibliometric analysis of federally sponsored studies 2000-2019

BACKGROUND

The rapid growth of inherently complex and heterogeneous data in HIV/AIDS research underscores the importance of Big Data Science. Recently, there have been increasing uptakes of Big Data techniques in basic, clinical, and public health fields of HIV/AIDS research. However, no studies have systematically elaborated on the evolving applications of Big Data in HIV/AIDS research. We sought to explore the emergence and evolution of Big Data Science in HIV/AIDS-related publications that were funded by the US federal agencies.

METHODS

We identified HIV/AIDS and Big Data related publications that were funded by seven federal agencies from 2000 to 2019 by integrating data from National Institutes of Health (NIH) ExPORTER, MEDLINE, and MeSH. Building on bibliometrics and Natural Language Processing (NLP) methods, we constructed co-occurrence networks using bibliographic metadata (e.g., countries, institutes, MeSH terms, and keywords) of the retrieved publications. We then detected clusters among the networks as well as the temporal dynamics of clusters, followed by expert evaluation and clinical implications.

RESULTS

We harnessed nearly 600 thousand publications related to HIV/AIDS, of which 19,528 publications relating to Big Data were included in bibliometric analysis. Results showed that (1) the number of Big Data publications has been increasing since 2000, (2) US institutes have been in close collaborations with China, Canada, and Germany, (3) some institutes (e.g., University of California system, MD Anderson Cancer Center, and Harvard Medical School) are among the most productive institutes and started using Big Data in HIV/AIDS research early, (4) Big Data research was not active in public health disciplines until 2015, (5) research topics such as genomics, HIV comorbidities, population-based studies, Electronic Health Records (EHR), social media, precision medicine, and methodologies such as machine learning, Deep Learning, radiomics, and data mining emerge quickly in recent years.

CONCLUSIONS

We identified a rapid growth in the cross-disciplinary research of HIV/AIDS and Big Data over the past two decades. Our findings demonstrated patterns and trends of prevailing research topics and Big Data applications in HIV/AIDS research and suggested a number of fast-evolving areas of Big Data Science in HIV/AIDS research including secondary analysis of EHR, machine learning, Deep Learning, predictive analysis, and NLP.

Social and Behavioral Determinants of Health in the Era of Artificial Intelligence with Electronic Health Records: A Scoping Review

Background. There is growing evidence that social and behavioral determinants of health (SBDH) play a substantial effect in a wide range of health outcomes. Electronic health records (EHRs) have been widely employed to conduct observational studies in the age of artificial intelligence (AI). However, there has been limited review into how to make the most of SBDH information from EHRs using AI approaches.Methods. A systematic search was conducted in six databases to find relevant peer-reviewed publications that had recently been published. Relevance was determined by screening and evaluating the articles. Based on selected relevant studies, a methodological analysis of AI algorithms leveraging SBDH information in EHR data was provided.Results. Our synthesis was driven by an analysis of SBDH categories, the relationship between SBDH and healthcare-related statuses, natural language processing (NLP) approaches for extracting SBDH from clinical notes, and predictive models using SBDH for health outcomes.Discussion. The associations between SBDH and health outcomes are complicated and diverse; several pathways may be involved. Using NLP technology to support the extraction of SBDH and other clinical ideas simplifies the identification and extraction of essential concepts from clinical data, efficiently unlocks unstructured data, and aids in the resolution of unstructured data-related issues.Conclusion. Despite known associations between SBDH and diseases, SBDH factors are rarely investigated as interventions to improve patient outcomes. Gaining knowledge about SBDH and how SBDH data can be collected from EHRs using NLP approaches and predictive models improves the chances of influencing health policy change for patient wellness, ultimately promoting health and health equity.

Novel Longitudinal Methods for Assessing Retention in Care: a Synthetic Review

PURPOSE OF REVIEW

Retention in care is both dynamic and longitudinal in nature, but current approaches to retention often reduce these complex histories into cross-sectional metrics that obscure the nuanced experiences of patients receiving HIV care. In this review, we discuss contemporary approaches to assessing retention in care that captures its dynamic nature and the methodological and data considerations to do so.

RECENT FINDINGS

Enhancing retention measurements either through patient tracing or "big data" approaches (including probabilistic matching) to link databases from different sources can be used to assess longitudinal retention from the perspective of the patient when they transition in and out of care and access care at different facilities. Novel longitudinal analytic approaches such as multi-state and group-based trajectory analyses are designed specifically for assessing metrics that can change over time such as retention in care. Multi-state analyses capture the transitions individuals make in between different retention states over time and provide a comprehensive depiction of longitudinal population-level outcomes. Group-based trajectory analyses can identify patient subgroups that follow distinctive retention trajectories over time and highlight the heterogeneity of retention patterns across the population. Emerging approaches to longitudinally measure retention in care provide nuanced assessments that reveal unique insights into different care gaps at different time points over an individuals' treatment. These methods help meet the needs of the current scientific agenda for retention and reveal important opportunities for developing more tailored interventions that target the varied care challenges patients may face over the course of lifelong treatment.

Development of a predictive model for retention in HIV care using natural language processing of clinical notes

OBJECTIVE

Adherence to a treatment plan from HIV-positive patients is necessary to decrease their mortality and improve their quality of life, however some patients display poor appointment adherence and become lost to follow-up (LTFU). We applied natural language processing (NLP) to analyze indications towards or against LTFU in HIV-positive patients' notes.

MATERIALS AND METHODS

Unstructured lemmatized notes were labeled with an LTFU or Retained status using a 183-day threshold. An NLP and supervised machine learning system with a linear model and elastic net regularization was trained to predict this status. Prevalence of characteristics domains in the learned model weights were evaluated.

RESULTS

We analyzed 838 LTFU vs 2964 Retained notes and obtained a weighted F1 mean of 0.912 via nested cross-validation; another experiment with notes from the same patients in both classes showed substantially lower metrics. "Comorbidities" were associated with LTFU through, for instance, "HCV" (hepatitis C virus) and likewise "Good adherence" with Retained, represented with "Well on ART" (antiretroviral therapy).

DISCUSSION

Mentions of mental health disorders and substance use were associated with disparate retention outcomes, however history vs active use was not investigated. There remains further need to model transitions between LTFU and being retained in care over time.

CONCLUSION

We provided an important step for the future development of a model that could eventually help to identify patients who are at risk for falling out of care and to analyze which characteristics could be factors for this. Further research is needed to enhance this method with structured electronic medical record fields.

Applying a machine learning modelling framework to predict delayed linkage to care in patients newly diagnosed with HIV in Mecklenburg County, North Carolina, USA

BACKGROUND

Machine learning has the potential to help researchers better understand and close the gap in HIV care delivery in large metropolitan regions such as Mecklenburg County, North Carolina, USA.

OBJECTIVES

We aim to identify important risk factors associated with delayed linkage to care for HIV patients with novel machine learning models and identify high-risk regions of the delay.

METHODS

Deidentified 2013-2017 Mecklenburg County surveillance data in eHARS format were requested. Both univariate analyses and machine learning random forest model (developed in R 3.5.0) were applied to quantify associations between delayed linkage to care (>30 days after diagnosis) and various risk factors for individual HIV patients. We also aggregated linkage to care by zip codes to identify high-risk communities within the county.

RESULTS

Types of HIV-diagnosing facility significantly influenced time to linkage; first diagnosis in hospital was associated with the shortest time to linkage. HIV patients with lower CD4+ cell counts (<200/ml) were twice as likely to link to care within 30 days than those with higher CD4+ cell count. Random forest model achieved high accuracy (>80% without CD4+ cell count data and >95% with CD4+ cell count data) to predict risk of delay in linkage to care. In addition, we also identified top high-risk zip codes of delayed linkage.

CONCLUSION

The findings helped public health teams identify high-risk communities of delayed HIV care continuum across Mecklenburg County. The methodology framework can be applied to other regions with HIV epidemic and challenge of delayed linkage to care.

Utilizing electronic health record data to understand comorbidity burden among people living with HIV: a machine learning approach

OBJECTIVES

An understanding of the predictors of comorbidity among people living with HIV (PLWH) is critical for effective HIV care management. In this study, we identified predictors of comorbidity burden among PLWH based on machine learning models with electronic health record (EHR) data.

METHODS

The study population are individuals with a HIV diagnosis between January 2005 and December 2016 in South Carolina (SC). The change of comorbidity burden, represented by the Charlson Comorbidity Index (CCI) score, was measured by the score difference between pre- and post-HIV diagnosis, and dichotomized into a binary outcome variable. Thirty-five risk predictors from multiple domains were used to predict the increase in comorbidity burden based on the logistic least absolute shrinkage and selection operator (Lasso) regression analysis using 80% data for model development and 20% data for validation.

RESULTS

Of 8253 PLWH, the mean value of the CCI score difference was 0.8 ± 1.9 (range from 0 to 21) with 2328 (28.2%) patients showing an increase in CCI score after HIV diagnosis. Top predictors for an increase in CCI score using the LASSO model included older age at HIV diagnosis, positive family history of chronic conditions, tobacco use, longer duration with retention in care, having PEBA insurance, having low recent CD4+ cell count and duration of viral suppression.

CONCLUSION

The application of machine learning methods to EHR data could identify important predictors of increased comorbidity burden among PLWH with high accuracy. Results may enhance the understanding of comorbidities and provide the evidence based data for integrated HIV and comorbidity care management of PLWH.

Power of Big Data in ending HIV

The articles in this special issue of AIDS focus on the application of the so-called Big Data science (BDS) as applied to a variety of HIV-applied research questions in the sphere of health services and epidemiology. Recent advances in technology means that a critical mass of HIV-related health data with actionable intelligence is available for optimizing health outcomes, improving and informing surveillance. Data science will play a key but complementary role in supporting current efforts in prevention, diagnosis, treatment, and response needed to end the HIV epidemic. This collection provides a glimpse of the promise inherent in leveraging the digital age and improved methods in Big Data science to reimagine HIV treatment and prevention in a digital age.

Using a machine learning approach to explore predictors of healthcare visits as missed opportunities for HIV diagnosis

OBJECTIVES

A significant number of individuals with a new HIV diagnosis are still late presenters despite numerous healthcare encounters prior to HIV diagnosis. We employed a machine learning approach to identify the predictors for the missed opportunities for earlier HIV diagnosis.

METHODS

The cohort comprised of individuals who were diagnosed with HIV in South Carolina from January 2008 to December 2016. Late presenters (LPs) (initial CD4 ≤200 cells/mm3 within one month of HIV diagnosis) with any healthcare visit during three years prior to HIV diagnosis were defined as patients with a missed opportunity. Using least absolute shrinkage and selection operator (LASSO) regression, two prediction models were developed to capture the impact of facility type (model 1) and physician specialty (model 2) of healthcare visits on missed opportunities.

RESULTS

Among 4,725 eligible participants, 72.2% had at least one healthcare visit prior to their HIV diagnosis, with most of the healthcare visits (78.5%) happening in the emergency departments (ED). A total of 1,148 individuals were LPs, resulting in an overall prevalence of 24.3% for the missed opportunities for earlier HIV diagnosis. Common predictors in both models included ED visit, older age, male gender, and alcohol use.

CONCLUSIONS

The findings underscored the need to reinforce the universal HIV testing strategy ED remains an important venue for HIV screening, especially for medically underserved or elder population. An improved and timely HIV screening strategy in clinical settings can be a key for early HIV diagnosis and play an increasingly important role in ending HIV epidemic.

Application of machine-learning techniques in classification of HIV medical care status for people living with HIV in South Carolina

OBJECTIVES

Ending the HIV epidemic requires innovative use of data for intelligent decision-making from surveillance through treatment. This study sought to examine the usefulness of using linked integrated PLWH health data to predict PLWH's future HIV care status and compare the performance of machine-learning methods for predicting future HIV care status for SC PLWH.

DESIGN

We employed supervised machine learning for its ability to predict PLWH's future care status by synthesizing and learning from PLWH's existing health data. This method is appropriate for the nature of integrated PLWH data because of its high volume and dimensionality.

METHODS

A data set of 8888 distinct PLWH's health records were retrieved from an integrated PLWH data repository. We experimented and scored seven representative machine-learning models including Bayesian Network, Automated Neural Network, Support Vector Machine, Logistic Regression, LASSO, Decision Trees and Random Forest to best predict PLWH's care status. We further identified principal factors that can predict the retention-in-care based on the champion model.

RESULTS

Bayesian Network (F = 0.87, AUC = 0.94, precision = 0.87, recall = 0.86) was the best predictive model, followed by Random Forest (F = 0.78, AUC = 0.81, precision = 0.72, recall = 0.85), Decision Tree (F = 0.76, AUC = 0.75, precision = 0.70, recall = 0.82) and Neural Network (cluster) (F = 0.75, AUC = 0.71, precision = 0.69, recall = 0.81).

CONCLUSION

These algorithmic applications of Bayesian Networks and other machine-learning algorithms hold promise for predicting future HIV care status at the individual level. Prediction of future care patterns for SC PLWH can help optimize health service resources for effective interventions. Predictions can also help improve retention across the HIV continuum.

Machine Learning and Clinical Informatics for Improving HIV Care Continuum Outcomes

PURPOSE OF REVIEW

This manuscript reviews the use of electronic medical record (EMR) data for HIV care and research along the HIV care continuum with a specific focus on machine learning methods and clinical informatics interventions.

RECENT FINDINGS

EMR-based clinical decision support tools and electronic alerts have been effectively utilized to improve HIV care continuum outcomes. Accurate EMR-based machine learning models have been developed to predict HIV diagnosis, retention in care, and viral suppression. Natural language processing (NLP) of clinical notes and data sharing between healthcare systems and public health agencies can enhance models for identifying people living with HIV who are undiagnosed or in need of relinkage to care. Challenges related to using these technologies include inconsistent EMR documentation, alert fatigue, and the potential for bias. Clinical informatics and machine learning models are promising tools for improving HIV care continuum outcomes. Future research should focus on methods for combining EMR data with additional data sources (e.g., social media, geospatial data) and studying how to effectively implement predictive models for HIV care into clinical practice.

Comorbidity patterns among people living with HIV: a hierarchical clustering approach through integrated electronic health records data in South Carolina

Comorbidity among people living with HIV (PLWH) is understudied although identifying its patterns and socio-demographic predictors would be beneficial for comorbidity management. Using electronic health records (EHR) data, 8,490 PLWH diagnosed between January 2005 and December 2016 in South Carolina were included in the current study. An initial list of 86 individual diagnoses of chronic conditions was extracted in the EHR data. After grouping individual diagnoses with a pathophysiological similarity, 24 diagnosis groups were generated. Hierarchical cluster analysis was applied to these 24 diagnosis groups and yielded four comorbidity clusters: "substance use and mental disorder" (e.g., alcohol use, depression, and illicit drug use); "metabolic disorder" (e.g., hypothyroidism, diabetes, hypertension, and chronic kidney disease); "liver disease and cancer" (e.g., hepatitis B, chronic liver disease, and non-AIDS defining cancers); and "cerebrovascular disease" (e.g., stroke and dementia). Multivariable logistic regression was conducted to investigate the association between socio-demographic factors and multimorbidity (defined as concurrence of ≥ 2 comorbidity clusters). The multivariable logistic regression showed that age, gender, transmission risk, race, initial CD4 counts, and viral load were significant factors associated with multimorbidity. The results suggested the importance of integrated clinical care that addresses the complexities of multiple, and potentially interacting comorbidities among PLWH.