In Journal of acquired immune deficiency syndromes (1999)
INTRODUCTION : Machine-learning (ML) algorithms are increasingly being utilised to inform HIV prevention and detection strategies. We validated and extended a previously developed ML model for patient retention on ART in a new geographic catchment area in South Africa.
METHODS : We compared the ability of an adaptive boosting (AdaBoost) algorithm to predict IIT in two South African cohorts from the Free State and Mpumalanga (FS/MP) and Gauteng and North West (GA/NW) provinces. We developed a novel set of predictive features for the GA/NW cohort using a categorical boosting model (CatBoost). We evaluated the ability of the model to predict IIT over all visits and across different time periods within a patient's treatment trajectory.
RESULTS : When predicting IIT, the GA/NW and FS/MP models demonstrated a sensitivity of 60% and 61% respectively, able to correctly predict nearly two thirds of all missed visits with a positive predictive value (PPV) of 18% and 19%. Using predictive features generated from the GA/NW cohort, the CatBoost model correctly predicted 22,119 of a total 35,985 missed next visits, yielding a sensitivity of 62%, specificity of 67% and PPV of 20%. Model performance was highest when tested on visits within the first six months.
CONCLUSIONS : Machine learning algorithms may be useful in informing tools to increase ART patient retention and efficiency of HIV care interventions. This is particularly relevant in developing countries where health data systems are being strengthened to collect data on a scale that is large enough to apply novel analytical methods.
Esra R, Carstens J, Le Roux S, Mabuto T, Eisenstein M, Keiser O, Orel E, Merzouki A, De Voux L, Maskew M, Sharpey-Schafer K