AI Model Can Accurately Assess PTSD in Women After Childbirth

A generative model of artificial intelligence (AI) that can analyze the narratives of women who have recently given birth has proven capable of accurately identifying complex post-traumatic stress disorder (CB-PTSD). This was revealed in a study by Massachusetts General Hospital (MGH), a founding member of the Mass General Brigham health system. By examining the capabilities and weaknesses of several OpenAI models, including ChatGPT, the researchers identified a version that provides rich insights into the mental health of mothers after a traumatic birth. The model can be seamlessly integrated into routine obstetric care and could potentially be used to assess other mental disorders as well. The results of the study were published in Scientific Reports.

How AI Detects Post-Traumatic Stress Disorder After Childbirth

“The assessment of PTSD related to traumatic childbirth currently relies on extensive clinical examinations, which do not meet the urgent need for a rapid and cost-effective assessment strategy,” says Dr. Sharon Dekel, director of the Postpartum Traumatic Stress Disorder Research Program at MGH and lead author of the study. “The use of brief patient narratives about childbirth, analyzed using text-based AI computational methods, may be an efficient, low-cost, and patient-friendly strategy for detecting CB-PTSD after traumatic childbirth. With further research, this tool could potentially help identify women at risk for such a disorder before the condition fully develops.”

For an estimated eight million women worldwide each year, a traumatic and/or medically complicated birth can trigger post-traumatic stress disorder, a condition historically associated with military combat or severe sexual assault. In recent years, childbirth has been recognized as a significant trigger for PTSD, which, if left untreated, can affect the health of both mother and child and result in significant societal costs. In previous studies, Dekel’s lab found evidence that brief psychological interventions shortly after a traumatic birth can reduce the mother’s birth-related PTSD symptoms.

In their latest study, Dekel, in collaboration with first author Alon Bartal, PhD, of Bar-Ilan University in Israel, investigated the effectiveness of artificial intelligence and related machine learning (ML) analysis methods for detecting CB-PTSD. Specifically, they evaluated the performance of various large language models (LLMs) and variants of ChatGPT, as well as their ability to gain new insights from text-based datasets derived from women’s brief narrative descriptions of their birth experience after giving birth. As part of their work, the team collected brief narrative reports from 1,295 women who had recently given birth. The study focused on an OpenAI model called “text-embeddings-ada-002,” which converted narrative data from the personal accounts of women with and without probable CB-PTSD into a numerical format that was then analyzed by a machine learning algorithm developed and trained by the team.

Cost-Effective Screening Strategy for Identifying Women at High Risk for PTSD

The researchers showed that this model outperforms other ChatGPT and large language models, which are typically trained on vast amounts of data to understand, analyze, and interpret natural language, in identifying postpartum traumatic stress. “The use of the ML model, which utilizes narrative inputs on the topic of childbirth from the OpenAI model as its sole data source, represents an efficient mechanism for data collection during the vulnerable postpartum period and exhibits 85 percent sensitivity and 75 percent specificity in identifying CB-PTSD cases,” Dekel notes. Furthermore, the model developed by the researchers could improve access to CB-PTSD screening and diagnosis, as it integrates seamlessly into routine obstetric care and provides a basis for commercial product development and widespread adoption. Dekel, whose research program is dedicated to studying the mental health of women after traumatic births, highlights the clinical benefits of using a pre-trained large language model to assess potential PTSD in young mothers. Early intervention is essential to prevent the disorder from progressing to a chronic stage, which can make treatment much more difficult.

This unique approach could introduce an innovative and cost-effective screening strategy to identify high-risk women and facilitate timely treatment. It could also show promise for assessing other mental disorders and, in turn, improving patient outcomes. The emergence of artificial intelligence tools in healthcare is groundbreaking and has the potential to positively transform continuity of care. Mass General Brigham, one of the leading integrated academic health systems and largest innovation companies in the US, is at the forefront of conducting rigorous research on new and emerging technologies to promote the responsible integration of AI into healthcare, workforce support, and administrative processes.