Background
In the UK, one in every 10,000 women dies during pregnancy or in the six weeks after giving birth. However, for every woman who dies, more than 100 experience severe complications that can result in lifelong health problems. Women who are Black, Asian or mixed race, and women living in deprived areas face disproportionately higher risks. Despite government commitments to reduce maternal mortality, rates have continued to rise, highlighting an urgent need for better prediction and prevention of pregnancy complications.
The PRiSMM (PReventing Severe Maternal Morbidity) data platform represents a transformative approach to understanding maternal health at a national scale, linking individual patient data from electronic health records across primary, secondary, and tertiary care via NHS Secure Data Environments. However, traditional clinical data capture only periodic snapshots of women’s health during pregnancy.
Novelty and Importance
This project will embed wearable technology into the PRiSMM platform, creating a unique opportunity to integrate continuous physiological monitoring with comprehensive clinical outcomes data. Wearable devices can capture heart rate, heart rate variability, respiratory rate, temperature, blood oxygen saturation, and physical activity patterns throughout pregnancy. However, robust evidence linking these digital biomarkers to clinically meaningful outcomes remains limited.
By applying advanced deep learning methods to multimodal data streams, this research will explore whether digital signatures can predict which women are at highest risk of severe maternal complications, potentially enabling earlier intervention and more equitable care.
Aims and Objectives
This PhD aims to explore digital signatures of health behaviours and predictors of pregnancy outcomes within the national PRiSMM dataset. Specific objectives are to:
(1) describe physical activity and physiological parameters captured via wearable devices across gestation;
(2) develop predictive models based on multimodal wearable and mobile device data to predict adverse health outcomes;
(3) validate these models in the PRiSMM dataset, with particular attention to performance across different demographic groups to ensure equitable prediction.
This project is co-funded by BiotrackOS, part of The Original Fit Factory.