Leveraging DHIS2 Data and Machine Learning for Maternal Mortality Prediction in Tanzania : A Comparative Study to Support Decision-Making
Maatu, Yusufu Hamisi (2024)
2024
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2024120933796
https://urn.fi/URN:NBN:fi:amk-2024120933796
Tiivistelmä
This study investigates ways to use machine learning (ML) techniques to improve the quality of maternal health data in Tanzania's DHIS2 system. Finding important risk factors that lead to maternal mortality and evaluating different machine learning methods were the main goals. By identifying intricate links within datasets, machine learning models including neural networks, support vector machines, and decision trees have been demonstrated to accurately predict maternal health outcomes (Brahimi et al., 2022; He et al., 2022; Zhang et al., 2023). "These models not only improve predictive accuracy but also provide actionable insights for targeted interventions, particularly in resource-constrained settings" (Zhou and colleagues, 2022).
The study demonstrates how machine learning (ML) may be applied to help guide policy choices, pinpoint important areas that need attention, and offer insightful information on maternal health statistics. It emphasizes how crucial it is to enhance data gathering and reporting systems, especially for important maternal health indicators, to facilitate well-informed decision-making. According to He et al. (2022), Brahimi et al. (2022), and Nyanjara et al. (2022), "It is essential to enhance the quality of data collection and reporting systems in order to improve maternal health outcomes and reduce maternal mortality in Tanzania."
The data showed that, in comparison to other algorithms, the decision tree model generated the most accurate forecasts with the lowest error rate, while linear regression yielded useful information, and random forest performed at a mediocre level (Nyanjara et al., 2022). Neural network and SVM models demonstrated low accuracy and high error rates, reflecting poor performance (He et al., 2022). The analysis identified that the three most important risk factors for maternal death were postpartum haemorrhage, eclampsia, and high blood pressure. However, discrepancies in the data related to conditions such as anemia and pre-eclampsia highlighted challenges with data quality (Brahimi et al., 2022).
The study demonstrates how machine learning (ML) may be applied to help guide policy choices, pinpoint important areas that need attention, and offer insightful information on maternal health statistics. It emphasizes how crucial it is to enhance data gathering and reporting systems, especially for important maternal health indicators, to facilitate well-informed decision-making. According to He et al. (2022), Brahimi et al. (2022), and Nyanjara et al. (2022), "It is essential to enhance the quality of data collection and reporting systems in order to improve maternal health outcomes and reduce maternal mortality in Tanzania."
The data showed that, in comparison to other algorithms, the decision tree model generated the most accurate forecasts with the lowest error rate, while linear regression yielded useful information, and random forest performed at a mediocre level (Nyanjara et al., 2022). Neural network and SVM models demonstrated low accuracy and high error rates, reflecting poor performance (He et al., 2022). The analysis identified that the three most important risk factors for maternal death were postpartum haemorrhage, eclampsia, and high blood pressure. However, discrepancies in the data related to conditions such as anemia and pre-eclampsia highlighted challenges with data quality (Brahimi et al., 2022).