Development of machine learning-based model for quality measurement in maternal, neonatal and child health services: a country level model for Tanzania

Abstract

Background: The high maternal and neonatal mortality in developing countries is frequently linked to inadequacies in the quality<br/>of maternal, neonatal and child health (MNCH) services provided. Quality measurement is among the recommended strategies for<br/>quality improvement in MNCH care. Consequently, developing countries require a novel quality measurement approach that can<br/>routinely facilitate the measurement and reporting of MNCH care quality. An effective quality measurement approach can<br/>enhance quality measurement and improve the quality of MNCH care. This study intends to explore the effectiveness of<br/>approaches available for MNCH quality measurement in developing countries. The study further proposes a machine learningbased<br/>approach for MNCH quality measurement.<br/>Method: A comprehensive literature search from Pub Med, HINARI, ARDI and Google Scholar electronic databases was<br/>conducted. Also, a search for organizations' websites, including World Health Organization (WHO), USAID's MEASURE<br/>Evaluation Project, Engender Health, and Family Planning 2020 (FP2020), was included. A search from databases yielded 324<br/>articles, 32 of which met inclusion criteria. Extracted articles were synthesized and presented.<br/>Findings: The majority of quality measurement approaches are manual and paper-based. Therefore are laborious, timeconsuming<br/>and prone to human errors. Also, it was observed that most approaches are costly since they require trained data<br/>collectors and special data sets for quality measurement. It is further noticed that the complexity of the quality measurement<br/>process and extra funds needed to facilitate data collection for quality measurement puts an extra burden on developing countries<br/>which always face constraints in health budgets. The study further proposes a machine learning-based approach for measuring<br/>MNCH quality. In developing this model, financial and human resource constrain were considered.<br/>Conclusion: The study found a variety of quality assessment approaches available for quality assessment on MNCH in developing<br/>countries. However, the majority of the existing approaches are relatively ineffective. Measuring MNCH quality by a machine<br/>learning-based approach could be advantageous and establish a much larger evidence base for MNCH health policies for<br/>Tanzania.