A comprehensive maternal health risk prediction dataset from IoT-enabled medical cyber-physical systems in developing countries: Supporting deep learning applications for clinical decision support

Background : This data article presents a comprehensive dataset of 6,103 maternal health records collected through Internet of Things (IoT)-enabled Medical Cyber-Physical Systems (MCPS) across multiple healthcare facilities in Bangladesh. Methods : The dataset comprises 8 attributes including clinical measurements (age, body temperature, heart rate, systolic and diastolic blood pressure, BMI, blood glucose levels) and risk classifications (high, mid, low risk). Data was collected using standardized IoT sensors including Raspberry Pi 4 controllers integrated with medical-grade sensors and validated by medical experts from 9 healthcare institutions between February 2021 and January 2023. Results : The dataset demonstrates balanced class distribution (Figure 1) and achieved up to 94.51% prediction accuracy using a Simple Recurrent Neural Network (RNN) with comprehensive cross-validation (Table 3). Temporal sequence modeling demonstrates strong performance, enabling enhanced interpretability through the use of attention mechanisms (Figure 9). Conclusions : This dataset addresses the critical need for maternal health risk prediction in resource-limited settings where traditional healthcare access is challenging. The data support the development of deep learning models, the creation of clinical decision support systems, and research on maternal mortality reduction. All data underwent rigorous quality control, expert validation, and ethical approval processes, making it suitable for academic research, healthcare technology development, and public health policy formulation in developing countries.