Machine Learning Predictions for HIV Patient Loss to Follow-Up in Haiti, 2018-2024

From 2018 to 2024, this dataset contains results from a study applying machine learning to predict loss to follow-up among 115,822 people living with HIV in Haiti. The analysis used national electronic medical record data from 167 health facilities, training models with demographic, clinical, and institutional predictors. The study was authored by Jiaqin Wu and shared under a CC-BY-4.0 license.

Use Cases

• Training predictive models for patient retention based on demographic, clinical, and institutional features mentioned in the description
• Identifying key risk factors for loss to follow-up using SHAP value analysis of model predictions
• Evaluating the performance of different machine learning algorithms (e.g., CatBoost) for time-series clinical forecasting
• Supporting the design of tiered intervention strategies based on individual risk scores derived from routine EMR data

Strengths

• Based on a large national cohort of 115,822 patients living with HIV
• Covers a six-year time range from 2018 to 2024 across 167 health facilities
• Model performance was rigorously assessed using multiple metrics including F2-score, recall, precision, ROC-AUC, and PR-AUC

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• The dataset is very small (21.3 KB), suggesting it may contain summary or model results rather than the underlying patient-level data

Provenance

Source

Haiti’s national electronic medical record database

Collection Method

Analysis of routine clinical data for patients receiving antiretroviral therapy

Time Range

2018 to 2024

Freshness

Last updated 2026-04-28 05:01:18; freshness should be verified

Geography

Haiti