Algorithmic prediction of HIV status using nation-wide electronic registry data

Abstract

a b s t r a c tBackground: Late HIV diagnosis is detrimental both to the individual and to society. Strategies to improve early diagnosis of HIV must be a key health care priority. We examined whether nation-wide electronic registry data could be used to predict HIV status using machine learning algorithms. Methods: We extracted individual level data from Danish registries and used algorithms to predict HIV status. We used various algorithms to train prediction models and validated these models. We calibrated the models to mimic different clinical scenarios and created confusion matrices based on the calibrated models. Findings: A total 4,384,178 individuals, including 4,350 with incident HIV, were included in the analyses. The full model that included all variables that included demographic variables and information on past medical history had the highest area under the receiver operating characteristics curves of 88 ·4% (95%CI: 87 ·5% -89 ·4%) in the validation dataset. Performance measures did not differ substantially with regards to which machine learning algorithm was used. When we calibrated the models to a specificity of 99 ·9% (pre-exposure prophylaxis (PrEP) scenario), we found a positive predictive value (PPV) of 8 ·3% in the full model. When we calibrated the models to a sensitivity of 90% (screening scenario), 384 individuals would have to be tested to find one undiagnosed person with HIV. Interpretation: Machine learning algorithms can learn from electronic registry data and help to predict HIV status with a fairly high level of accuracy. Integration of prediction models into clinical software systems may complement existing strategies such as indicator condition-guided HIV testing and prove useful for identifying individuals suitable for PrEP.

Machine learning to identify socio-behavioural predictors of HIV positivity in East and Southern Africa

“…This methodology has been implemented to establish patterns of HIV risk behaviour, to optimise HIV treatment modalities and, to identify high-risk individuals for targeted interventions from a number of novel data sources. 6–15…”

Machine Learning Classification of False-Positive Human Immunodeficiency Virus Screening Results

“…[ 10 11 12 13 14 15 16 17 18 ] There has been limited work using EMR demographic data to predict HIV status which could be used by health-care systems to proactively encourage screening. [ 19 20 21 ] One group has applied machine learning to laboratory testing, by using flow cytometry images of CD4+ cells to assess diagnosis and treatment of HIV,[ 18 ] providing valuable information, but not creating a decision support tool for the clinical laboratory.…”

Technology-Delivered Intervention Strategies to Bolster HIV Testing

“…Seven testers felt that the interaction was private and anonymous and therefore, they would be willing to take a test with the chatbot. Vermey et al, 2019 [ 58 ] Assessment of an online advice application with a chatbot feature that provides tailored advice based on user-provided information regarding personal characteristics, sexual behavior, sexual risk, and symptoms. The online application was visited 337,736 times in 2018 with 113,257 visitors started the questionnaire, and 17,449 utilizing the chatbot.…”

“…Through this process, investigators were able to identify acute seroconverters, new chronically ill patients, and known infected patients (many who did not disclose their status) [ 56 ]. Investigators utilized machine learning algorithms that would identify patients at risk for HIV acquisition among adult members of a large healthcare organization [ 57 ] and national hospital registry [ 58 ]. In both studies, various models were tested to discriminate and correctly predict between HIV cases and non-cases.…”