Monitoring system for pregnant women with IoT and AI tools: Case study: San Pedro Hospital Foundation - Pasto, Colombia

Abstract

Extreme Maternal Morbidity (EMM) is a critical public health issue in Colombia, particularly in Nariño, where socioeconomic factors and barriers to medical care have increased its incidence. To address this challenge, a monitoring system based on Artificial Intelligence (AI) and the Internet of Things (IoT) has been developed at Fundación Hospital San Pedro (FHSP) in Pasto, aiming to reduce severe obstetric complications through early detection and continuous monitoring.

This project responds to the need to optimize the identification and follow-up of patients at risk of EMM. Its objectives include: (i) the creation of an AI-based risk classification model, categorizing patients into High, Medium, and Low risk; (ii) the development of a mobile application for data collection and clinical analysis at FHSP; (iii) the implementation of an IoT Kit for extramural monitoring of high-risk patients; and (iv) the structuring of a prevention program based on the collected data.

The methodology involves analyzing 21,770 patient records from FHSP between 2013 and 2023, applying Machine Learning models such as Logistic Regression, Random Forest, XGBoost, and Neural Networks, achieving over 98% accuracy. This analysis enabled the design of a mobile application that classifies obstetric risk in real time, improving clinical follow-up. In parallel, the IoT Kit measures blood pressure, heart rate, oxygen saturation, and temperature, enabling remote monitoring of high-risk patients.

The results include two key tools: (1) a mobile application currently in use at FHSP, facilitating data collection and risk alert generation, and (2) an IoT Kit, currently in the testing phase, which transmits real-time data to a cloud-based platform for continuous patient monitoring. Additionally, the integration of these developments has led to the establishment of a prevention program, providing critical epidemiological data for the formulation of maternal health policies.

This innovative system combines the predictive power of Machine Learning with the connectivity and real-time monitoring capabilities of IoT, offering a scalable solution to enhance prenatal care and reduce maternal mortalityin vulnerable populations.