Advancing Maternal Health Through AI-Powered Gestational Diabetes Prediction

GDM is a prevalent and growing health issue globally, with adverse effects on maternal and neonatal health. However, there is a lack of specific AI tools trained on GDM data from the South Asian population, including Pakistan, which hinders early prediction and intervention efforts. Furthermore, comprehensive and region-specific data are scarce on GDM prevalence, risk factors, and outcomes in South Asian populations, including countries like Pakistan. This knowledge gap hinders the development of targeted interventions and preventive measures. The existing research and predictive models often rely on data from different populations, which may not fully capture the specific risk factors and characteristics of GDM in South Asia. Addressing this knowledge gap is essential to develop more effective and accurate tools for early prediction and intervention of GDM in this region.

This research will play a vital role in improving maternal and neonatal health outcomes, reducing the burden of GDM-related complications, and guiding healthcare providers in managing GDM effectively in the South Asian population. Pakistan’s demographic landscape is characterized by significant diversity. Our medical partner in this project, Shifa International, possesses the expertise and capability to cater to a wide spectrum of individuals, encompassing both rural and urban populations, as well as people from various demographic backgrounds. The target population will generally consist of rural and urban populations from ICT, Punjab and AJK region (being major clientele of Shifa International Hospital). However, being a major tertiary care hospital, it is expected that a substantial number of cases from across the country will also be included.

Furthermore, a robust and industry-standard complaint Hospital Management and Information System (HMIS) is deployed in Shifa International to capture high-quality data which will be used to define the prediction models. It is important to highlight that our approach involves close collaboration with a medical team actively engaged in dealing with pregnant women having GDM, along with their entire clinical and social history (as recorded by the hospital as a standard practice). Their firsthand experiences will be continuously integrated into our model development and software design process from the project’s outset, ensuring that women’s unique needs and challenges are addressed comprehensively throughout the research.

AIM

To implement an AI-based GDM screening and outcome prediction platform for a diverse demographic population of South Asia by selecting Pakistan as a base case study.

OBJECTIVES

• Develop an evidence-based AI-powered platform for GDM risk prediction in expectant mothers, utilizing Electronic Medical Records (EMR) from Shifa International Hospital

• Establish a data extraction pipeline, gathering comprehensive clinical data, selecting pertinent features, and constructing a dependable predictive model

• Enhance eShifa platform  to provide real-time GDM risk assessments and personalized alerts and notifications to empower expectant mothers in managing their health

• Predict maternal and fetal outcomes based on the result of GDM prediction