Researchers aim to ensure future AI healthcare monitoring systems are free of gender bias

The project will train a newly developed AI system on data collected from participating men and women

Researchers from the University of Glasgow’s James Watt School of Engineering are aiming to ensure that artificial intelligence (AI)-based healthcare monitoring systems in the future are free of gender bias, to improve care for both men and women.

For 18 months, the project will examine the potential for gender bias in healthcare AI and discover ways to ensure that AI-supported treatment remains equitable.

The use of cutting-edge sensors is currently being investigated to track the rhythms of patients’ hearts and lungs without requiring them to wear monitoring devices or be recorded on video cameras.

The team aims to address and ensure that its AI component is properly trained and capable of making the correct judgements without bias towards one gender of patients.

Supported by £8,200 in funding from the Université Paris Dauphine-PSL’s Women and Science Chair, with further support from the L’Oréal Foundation, Generali France, La Poste, Amundi and the Talan Group, researchers will work to develop a new framework to balance gender-related behaviours in an AI monitoring system.

Researchers will collect healthcare data from 30 male and 30 female study volunteers using radar sensors to be used to train a newly developed AI architecture and analyse the results of the radar monitoring.

Additionally, separate models will be trained using the male and female data to compare performance and highlight any biases in AI’s performance to be adjusted for using statistical models and mitigation techniques.

The vital sign monitoring systems will work to identify the signs of an unexpected change in heart rate or respiration in a less-invasive format. If identified, the system will send an alert for help.

Project supervisor Dr Julien Le Kernec, James Watt School of Engineering, commented: “By the end of our research,… we hope the outputs from this project will help guide the future development of this… technology and ensure that the AI which supports it is fair and balanced for the many diverse groups of patients who will benefit from it.”