Developed by AI and data scientist Professor Georgina Cosma and human factors and complex systems expert Professor Patrick Waterson, the tool analyses maternity incident reports to highlight key human factors – such as communication, teamwork, and decision-making – that may have impacted care outcomes, providing insights into areas that could benefit from additional support.
Currently, experts must carry out manual reviews to extract human factor insights from incident reports. This process is resource-intensive, time-consuming, and relies on individual interpretation and expertise, which can lead to varying conclusions.
The AI tool addresses these challenges by identifying and categorising human factors in reports quickly and consistently. Its standardised approach allows it to analyse multiple reports and identify recurring factors, helping pinpoint areas that would benefit most from additional support.
Prof Waterson said: ‘The need for such research was highlighted in the Ockenden Review, which examined maternity care and set to improve safety and care quality in maternity services.
"By taking a more comprehensive view of maternal healthcare delivery, we can develop targeted interventions to improve maternal outcomes for all mothers and babies.'
The AI model was trained and tested on data from 188 real maternity incident reports. It successfully identified human factors in each report and analysed them collectively, providing insights into where extra support could improve outcomes.
Prof Cosma said: ‘AI has transformed our analysis of maternity safety reports. We've uncovered crucial insights far quicker than manual methods.
‘This has enabled us to gather a comprehensive understanding of where there are areas for improvement in maternity care, and these insights can help identify ways to enhance patient safety and improve outcomes for mothers and babies.'
Teamwork and communication emerged as the most frequently identified human factors across all the analysed reports.
The analysis also emphasised the importance of thorough patient evaluations – including assessments, investigations, and screenings – as well as the impact of individual patient characteristics, such as birth history and conditions like pre-eclampsia, on care outcomes.
The AI tool identified challenges related to medical technology use and staff performance, indicating that ongoing training and support could improve care outcomes. It also provided insights into how Covid-19 affected maternity services, underscoring the need for adaptability in practices.
The analysis also indicated that certain human factors might have a greater impact on mothers from ethnic minority groups. However, due to the limited number of reports that included ethnicity data, further research is required to reach definitive conclusions.
The Loughborough researchers hope to secure funding to refine the AI model using a larger, more diverse dataset as expanded testing is essential to validate the tool's effectiveness and further understand the challenges faced by mothers from ethnic minority groups in maternity care.
Prof Cosma added: ‘We are seeking to collaborate with hospitals, healthcare organisations, and investigation bodies to further refine and apply our AI tool to reports. These partnerships will help us extract vital intelligence to prevent adverse incidents and ensure the safety of all mothers and babies.'
In response, Dr Jonathan Back, a safety insights analyst at The Health Services Safety Investigations Body, said the tool ‘could help analysts working in health and care to identify where there are inequalities, maximising learning by bringing together findings from multiple investigations'.
