Dataset Information

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

ABSTRACT: Background: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. Funding:The study was supported by funds from the Preben and Anne Simonsens Foundation, the Novo Nordisk Foundation, Rigshospitalet, Copenhagen University, the Danish AIDS Foundation, the Augustinus Foundation and the Danish Health Foundation.

SUBMITTER: Ahlstrom MG

PROVIDER: S-EPMC6933258 | biostudies-literature | 2019 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

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

Ahlström Magnus G MG Ronit Andreas A Omland Lars Haukali LH Vedel Søren S Obel Niels N

EClinicalMedicine 20191105

<h4>Background</h4>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.<h4>Methods</h4>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. W ...[more]

PMID: 31891137

Similar Datasets

Project description:ContextImproving retention in care is a key element of the National HIV/AIDS Strategy (NHAS). However, definitions for measuring retention in care are not standardized.ObjectiveTo compare measures of retention based on both clinic visit data and HIV laboratory surveillance data.DesignRetrospective cohort study.SettingNew York City (NYC), New York.ParticipantsWe matched adult patients with HIV infection seen at the Spencer Cox Center for Health (SCC) in 2010 or 2011 with the NYC HIV Surveillance Registry.Main outcome measuresRetention in care was measured on the basis of SCC electronic medical record (EMR) data (?1 medical visits in 2012) and Surveillance Registry data (?2 CD4/viral load [VL] tests ?90 days apart in 2012).ResultsThere were 5746 adult HIV-infected patients seen at SCC between 2010 and 2011 who matched with the Surveillance Registry. Seventy-eight percent (n = 4469) had 1 or more medical visits at SCC in 2012 and were considered retained on the basis of the EMR definition, among which 3831 (86%) met the surveillance definition for retention in care. Patients who did not have a medical visit at SCC in 2012 (n = 1277) were lost to care in NYC (n = 485; 36%), engaged in care at an alternate provider (n = 622; 49%), or died after their last SCC visit (n = 197; 15%).ImplicationsThis study is an important comparison of laboratory surveillance versus clinic visit-based measures of retention in care in an urban setting with the largest HIV epidemic in the country. Collaborative projects between local health departments and clinical care providers can help validate the care status of patients and inform the allocation of resources to reengage patients who are lost to care.ConclusionThe combined use of laboratory and clinic visit-based data to measure retention in care provides a more accurate representation of the care status of HIV-infected patients than use of a single data source alone. Routine sharing of data by public health institutions and clinical care providers would help target resources toward reengaging patients who are lost to care in jurisdictions with universal HIV-related laboratory reporting.

Dataset Information

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

Publications

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets