AI found to rapidly, accurately identify people at risk of postpartum haemorrhage

They found the AI model to be 95 per cent accurate in identifying patients with the condition, and also identified 47 per cent more patients than when using the standard method.

New Delhi: Artificial intelligence (AI) could help clinicians identify people at a higher risk of postpartum haemorrhage, a common pregnancy complication, researchers have found.

The researchers at Brigham and Women's Hospital, US, used a large language model Flan-T5 to extract medical concepts from electronic health records to better define and identify the populations impacted by postpartum haemorrhage, which involves excessive bleeding after childbirth and can be life-threatening.

A large language model is trained on massive amounts of textual data and can thus process, manipulate and generate textual content.

They found the model to be 95 per cent accurate in identifying patients with the condition, and also identified 47 per cent more patients than when using the standard method. Their study is published in the journal npj Digital Medicine.

"We need better ways to identify the patients that have this complication, as well as the different clinical factors associated with it," said corresponding author Vesela Kovacheva of the Department of Anesthesiology, Perioperative and Pain Medicine at the hospital.

The research team used the Flan-T5 model to extract concepts related to postpartum haemorrhage from the discharge summaries of a cohort of more than 1.3 lakhs patients who gave birth at Mass General Brigham hospitals between 1998-2015. They achieved this by prompting the language model with lists of concepts known to be associated with the medical condition.

The researchers found this method to achieve rapid and accurate results.

'We looked at all of the patients that Flan-T5 identified as having postpartum haemorrhage and looked at what fraction of those also had the corresponding billing code. It turns out that Flan-T5 was 95 per cent accurate and allowed us to identify 47 per cent more patients than we would have from the billing codes alone,” said first author Emily Alsentzer, a research fellow in the Division.

"Ideally, we would like to be able to predict who will develop postpartum haemorrhage before they do so, and this is a tool that can help us get there," said Alsentzer.

The researchers plan to continue using this approach to look at other pregnancy complications.

'This approach can be applied to many future studies,' said Kovacheva. 'And it could be used to help guide real-time medical decision making, which is very exciting and valuable to me as a clinician.' PTI KRS SAR KRS