Journal of acquired immune deficiency syndromes (1999) · 2022
Predictive Analytics Using Machine Learning to Identify ART Clients at Health System Level at Greatest Risk of Treatment Interruption in Mozambique and Nigeria.
СтатьяПоддержка правительства СШАПоддержка службы здравоохранения СШАНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Stockman Jeni, Friedman Jonathan, Sundberg Johnna, Harris Emily, Bailey Lauren
Аннотация
BACKGROUND: A core objective of HIV/AIDS programming is keeping clients on treatment to improve their health outcomes and to limit spread. Machine learning and artificial intelligence can combine client, temporal, and locational attributes to identify which clients are at greatest risk of loss to follow-up (LTFU) and enable health providers to direct support interventions accordingly.
SETTING: The analysis was part of a project funded by U.S. President's Emergency Plan for AIDS Relief and United States Agency for International Development, Data for Implementation, and applied to data from publicly available sources (health facility data, geospatial data, and satellite imagery) and de-identified electronic medical record data on antiretroviral therapy clients in Nigeria and Mozambique.
METHODS: The project applied binary classification techniques using temporal cross-validation to predict the risk that patients would be LTFU. Classifiers included logistic regression, neural networks, and tree-based models.
RESULTS: Models showed strong predictive power in both settings. In Mozambique, the best-performing model, a Random Forest, achieved an area under the precision-recall curve of 0.65 compared against an underlying LTFU rate of 23%. In Nigeria, the best-performing model, a boosted tree, achieved an area under the precision-recall curve of 0.52 compared against an underlying LTFU rate of 27%.
CONCLUSIONS: Machine-learned models outperformed current classification techniques and showed potential to better direct health worker resources toward patients at greatest risk of LTFU. Moreover, models performed equally across sex and age groups, supporting the model's generalizability and wider application.
SETTING: The analysis was part of a project funded by U.S. President's Emergency Plan for AIDS Relief and United States Agency for International Development, Data for Implementation, and applied to data from publicly available sources (health facility data, geospatial data, and satellite imagery) and de-identified electronic medical record data on antiretroviral therapy clients in Nigeria and Mozambique.
METHODS: The project applied binary classification techniques using temporal cross-validation to predict the risk that patients would be LTFU. Classifiers included logistic regression, neural networks, and tree-based models.
RESULTS: Models showed strong predictive power in both settings. In Mozambique, the best-performing model, a Random Forest, achieved an area under the precision-recall curve of 0.65 compared against an underlying LTFU rate of 23%. In Nigeria, the best-performing model, a boosted tree, achieved an area under the precision-recall curve of 0.52 compared against an underlying LTFU rate of 27%.
CONCLUSIONS: Machine-learned models outperformed current classification techniques and showed potential to better direct health worker resources toward patients at greatest risk of LTFU. Moreover, models performed equally across sex and age groups, supporting the model's generalizability and wider application.