IIM Visakhapatnam Researchers Create MMR Dashboard Using Machine Learning

New Delhi: Researchers from the Indian Institute of Management, IIM Visakhapatnam have developed a dashboard/decision support system (DSS) to address maternal mortality. By utilizing advanced and interpretable machine learning models, they analyzed data from NFHS-4, NFHS-5, and the government's health management information system (HMIS).

This innovative dashboard developed by IIM Visakhapatnam researchers will provide targeted healthcare interventions for administrative units/districts to effectively reduce maternal mortality rates (MMR).

Recognizing the deep insights and performance evaluation capabilities of the DSS, IIM Visakhapatnam is in discussions with authorities to expand its benefits nationwide for the greater well-being of society. 

IIM Visakhapatnam DSS Highlights 

• IIM Visakhapatnam established the 'Interdisciplinary Decision Science and Analytics Lab' (IDeAL), which is a Centre of Excellence for inter-disciplinary research.
• IDeAL initially developed a dashboard for monitoring the Covid-19 pandemic.
• Researchers at IDeAL have now created a dashboard and decision support system focusing on maternal mortality.
• The MMR dashboard utilizes explainable machine learning models based on NFHS4, NFHS5, and HMIS data.
• It provides healthcare intervention recommendations specific to administrative units.
• The dashboard considers various factors such as healthcare infrastructure, topography, agroclimatic conditions, and insurance when suggesting interventions.
• Policymakers can utilize the DSS to modify existing interventions or customize targeted interventions.

Dr. Shivshanker Singh Patel, the creator of the dashboard and the head of the Inter-disciplinary Decision Science and Analytics Lab (IDeAL) at IIM Vizag, explained that maternal health is a critical global concern, particularly in developing nations.

He mentioned that in a country like India, which has a large population, there are significant differences in maternal health services and maternal mortality rates among different states. A targeted and precise healthcare strategy is required to effectively address this challenge.

Further, Dr. Patel revealed that the DSS dashboard runs on around 30-35 variables, including factors like medical insurance, agro climatic conditions, healthcare accessibility, geography, nutrition, healthcare facilities, and literacy levels.

According to Dr. Patel, various explainable AI models were utilized to analyze a wide range of factors associated with MMR and generate insights and recommendations. He explained that one of the models used was CART heuristics, which helped categorize districts into lower and higher MMR classes.

Another notable model employed was Shapley additive explanations, which incorporated machine learning techniques like support vector machines, artificial neural networks, boosting, and random forest to identify regions with different MMR levels.

An IIM Vizag Faculty member in the decision science area at IIM Vizag said that explainable boosting machines were also used, and the results were carefully compared to provide valuable policy suggestions.

Dr. Shivshanker Singh Patel, an experienced machine learning expert, highlighted that policymakers have the opportunity to enhance maternal health outcomes by tailoring interventions and policies to the specific needs of different regions.

This can be achieved by considering the precise recommendations and insights at the administrative unit level. He further added that the integration of precision healthcare strategies and explainable machine learning techniques holds great potential in addressing the complexities of maternal health disparities.

Thus, by focusing on personalized interventions and evidence-based policymaking, significant progress can be made in improving maternal health and reducing maternal mortality rates.

Read More:

• IIM Visakhapatnam Admission 2023 

• IIM Visakhapatnam Courses

Subscribe to Collegedunia to get the latest educational news and updates –