AI model can accurately screen for post-traumatic stress disorder after childbirth

A generative artificial intelligence (AI) model that can analyze the stories of women who have recently given birth can accurately screen for post-traumatic stress disorder (CB-PTSD), a Massachusetts General Hospital (MGH) study shows. Discovered by founding members of the Massachusetts General Brigham Health System.

By exploring the features and shortcomings of several models in OpenAI, including ChatGPT, researchers identified a version that provides rich insight into mothers' mental health after traumatic childbirth.

This model can be seamlessly adapted to routine obstetric care and may also be used to assess other mental health disorders. The research results are scientific report.

Assessment of PTSD related to traumatic childbirth currently relies on extensive clinician assessment, which does not meet the urgent need for rapid and low-cost assessment strategies. ”

Dr. Sharon Dekel, director of MGH's Postpartum Traumatic Stress Disorder Research Program and senior author of this study

“The use of short patient narratives about childbirth analyzed by AI text-based computational methods could be an efficient, low-cost, and patient-friendly strategy for detecting CB-PTSD after traumatic childbirth.” Yes, and with further research, this tool could potentially help treat: “Identifying women at risk for CB-PTSD before the condition has fully developed.'' It is to do. ”

For an estimated 8 million women annually worldwide, traumatic and/or medically complicated births are expected to result in post-traumatic stress disorder. Post-traumatic stress disorder has historically been associated with military combat and severe sexual assault.

In recent years, childbirth has been recognized as a significant PTSD trigger, which, if left untreated, can impair the health of both mother and child and result in significant societal costs.

In previous research, Dekel's lab found evidence that brief psychological interventions immediately following a traumatic birth can reduce mothers' birth-related PTSD symptoms.

In the latest study, Dekel collaborated with lead author Dr. Alon Bartal of Bar-Ilan University in Israel to investigate the effectiveness of artificial intelligence and related machine learning (ML) analysis strategies for detecting CB-PTSD. Did.

Specifically, we evaluate the performance of different large-scale language models (LLMs) and ChatGPT variations, as well as their ability to extract new insights from a text-based dataset derived from postpartum women's short narrative descriptions of their childbirth experiences. Did.

As part of the study, the team collected short stories from 1,295 women who had just given birth.

This research focused on an OpenAI model known as text-embeddings-ada-002. The model converted narrative data from personal accounts of women with and without possible CB-PTSD into numerical form, which was then analyzed by a trained machine learning algorithm developed by . team.

Researchers showed that this model performed better in identifying postpartum traumatic stress compared to other ChatGPT and large-scale language models. These models are typically trained on vast amounts of data that enable them to understand, analyze, and interpret natural language.

“Reliance on ML models that use birth narrative input from Open AI models as a dedicated data source provides an efficient mechanism for data collection during the vulnerable period of the puerperal period and is highly effective in identifying CB-PTSD cases 85 percent sensitivity and 75 percent specificity,” Dekel noted.

“Furthermore, the model we developed fits seamlessly into routine obstetric care and provides the foundation for commercial product development and mainstream adoption, improving access to CB-PTSD screening and diagnosis.” There is likely to be.”

Dr. Dekel's research program specializes in investigating women's mental health after traumatic childbirth, using large-scale, pre-trained language models to identify potential PTSD in new mothers. highlights the clinical benefits of evaluating

“Early intervention is essential to prevent the disease from progressing to a chronic stage, which can be very complex to treat,” MGH researchers note.

“Our unique approach has the potential to introduce innovative and cost-effective screening strategies to identify high-risk women and facilitate timely treatment. It also holds promise for assessing health disorders and improving patient outcomes as a result.”

The emergence of artificial intelligence tools in healthcare is revolutionary and has the potential to positively reshape the continuum of care. As one of the nation's top integrated academic health systems and largest innovation companies, Mass General Brigham is seeking information on how to responsibly incorporate AI into care delivery, workforce support, and administrative processes. We lead rigorous research into new technologies.

Dekel is a psychologist at MGH and an assistant professor of psychology at Harvard Medical School. Bartal is an assistant professor of information systems at Bar-Ilan University in Israel. Co-authors from the Dekel Institute include Harvard researcher Dr. Kathleen Jagodnik and clinical research coordinator Sabrina Chan.

Dekel was supported by funding from the NIH (Eunice Kennedy Shriver National Institute of Child Health and Human Development, grants R01HD108619, R21HD109546, and R21HD100817).