This community innovation has been accepted at the 2026 DHIS2 Annual Conference and will be in abstract track/lightning talk.
Predictive Analytics in DHIS2 for ART Default Risk
Background Antiretroviral treatment (ART) default among adolescents and young people living with HIV remains a persistent challenge in urban settings such as Harare Province, despite Zimbabwe’s progress toward the UNAIDS 95–95–95 targets. Existing monitoring approaches, including DHIS2 and electronic patient monitoring systems, operate reactively by identifying defaulters after missed clinic visits, limiting timely interventions and increasing risks of viral rebound, drug resistance, and loss to follow-up. Objective To develop and evaluate a predictive analytics approach using routinely collected DHIS2-compatible HIV program data to assess ART default risk among young people aged 10–24 years in Harare Province and demonstrate operational use within DHIS2. Description A retrospective cohort study analysed 17,056 records of young people receiving HIV care in Harare Province between May 2024 and April 2025. The Cross-Industry Standard Process for Data Mining (CRISP-DM) guided analysis. Data preprocessing included cleaning, median imputation, feature scaling and class balancing using SMOTE. Predictor variables included demographics, clinic attendance, viral load suppression, CD4 history, ART regimen, and duration on ART. Supervised machine learning models were developed including logistic regression, random forest, XGBoost, support vector machines, and a recurrent neural network (RNN). The RNN added value by modelling longitudinal care trajectories and generating continuous ART default risk scores, categorised into low, medium and high risk and re-integrated into DHIS2 dashboards. Lessons Learnt ART default among young people follows identifiable clinical and behavioural patterns particularly viral suppression and recent clinic attendance. Routine DHIS2-compatible data supported accurate prediction and improved dashboard-based program ownership. Conclusion DHIS2-based predictive analytics can strengthen youth ART retention through interventions.
Primary Author: MIRIRAI MUYENGWA
Keywords:
DHIS2, antiretroviral therapy (ART) default; predictive analytics,machine learning, adolescents and young people living with HIV,ART retention
1 Like