Dataset Information

AI and narrative embeddings detect PTSD following childbirth via birth stories.

ABSTRACT: Free-text analysis using machine learning (ML)-based natural language processing (NLP) shows promise for diagnosing psychiatric conditions. Chat Generative Pre-trained Transformer (ChatGPT) has demonstrated preliminary initial feasibility for this purpose; however, whether it can accurately assess mental illness remains to be determined. This study evaluates the effectiveness of ChatGPT and the text-embedding-ada-002 (ADA) model in detecting post-traumatic stress disorder following childbirth (CB-PTSD), a maternal postpartum mental illness affecting millions of women annually, with no standard screening protocol. Using a sample of 1295 women who gave birth in the last six months and were 18+ years old, recruited through hospital announcements, social media, and professional organizations, we explore ChatGPT's and ADA's potential to screen for CB-PTSD by analyzing maternal childbirth narratives. The PTSD Checklist for DSM-5 (PCL-5; cutoff 31) was used to assess CB-PTSD. By developing an ML model that utilizes numerical vector representation of the ADA model, we identify CB-PTSD via narrative classification. Our model outperformed (F1 score: 0.81) ChatGPT and six previously published large text-embedding models trained on mental health or clinical domains data, suggesting that the ADA model can be harnessed to identify CB-PTSD. Our modeling approach could be generalized to assess other mental health disorders.

SUBMITTER: Bartal A

PROVIDER: S-EPMC11009279 | biostudies-literature | 2024 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

AI and narrative embeddings detect PTSD following childbirth via birth stories.

Bartal Alon A Jagodnik Kathleen M KM Chan Sabrina J SJ Dekel Sharon S

Scientific reports 20240411 1

Free-text analysis using machine learning (ML)-based natural language processing (NLP) shows promise for diagnosing psychiatric conditions. Chat Generative Pre-trained Transformer (ChatGPT) has demonstrated preliminary initial feasibility for this purpose; however, whether it can accurately assess mental illness remains to be determined. This study evaluates the effectiveness of ChatGPT and the text-embedding-ada-002 (ADA) model in detecting post-traumatic stress disorder following childbirth (C ...[more]

PMID: 38605073

Similar Datasets

Project description:IntroductionInvolving service users in health service design and delivery is considered important to improve the quality of healthcare because it ensures that the delivery of healthcare is adapted to the needs of the users. Co-production is a process used to involve service users, but multiple papers have highlighted the need for the mechanisms and values guiding co-production to be more clearly stated. The aim of this paper was to evaluate the mechanisms and values that guided the co-production approach of the Stories for Change project, which used Public Narrative as part of the co-design process to create change in National Health Service maternity services.MethodsThis study was conducted using a rapid feedback evaluation approach. Semistructured interviews (n = 16) were the main source of data, six of which were maternity service users, with observations (5 h) and documentary analysis also carried out in parallel. RREAL sheets were used for data analysis to organize data based on key topics of interest.ResultsThis study identified three broad mechanisms and values underpinning the co-production approach: creating an open and safe space to share ideas, learning how to tell stories using Public Narrative and having service providers who play a key role in strengthening the health system listen to stories compelling them to action. This study identified the main areas for improvement of the Stories for Change project related to recruitment, the inclusion of participants, the co-design process, the Skills Session and the Learning Event.ConclusionOur study provided a deeper understanding of the co-production approach that addresses the need to uncover the mechanism and values underlying co-production and co-design approaches. This study expands on the literature pertaining to the influence of storytelling in creating meaningful change in health care. We propose a co-design methodology that uses Public Narrative as a model for service user engagement to help inform future healthcare development processes.Patient or public contributionThe experiences and perceptions of maternity service users and health professionals informed this evaluation. The project organizers were involved in the manuscript preparation stage by providing feedback, and service users wrote a commentary on the project from the lived experience perspective.

Project description:Background and aimRates of cesarean section in Iran are unnecessarily high largely due to fear of childbirth (FOC), yet this may be reduced through education. Iranian women are keen to obtain information about pregnancy and birth online though sources may not be reliable. Consequently, the present study aimed to compare the effect of childbirth preparation courses delivered both online via the social media platform 'Telegram' and in-person on pregnancy experience, FOC, birth preference, and mode of birth.MethodsThis quasi-experimental study included 165 primiparous pregnant women referred to the prenatal clinic in Tehran, Iran. Convenience sampling was used to recruit participants, who were subsequently divided into three groups; (A) social media-based educational intervention (n = 53); (B) in-person educational intervention (n = 52), and (C) a control group who received no prenatal education (n = 50). During the 18th and 20th weeks of pregnancy, demographic questions along with the pregnancy experience scale (PES), and version A of the Wijma delivery expectancy/experience questionnaire (WDEQ‑A) were completed. In the 36th and 38th weeks of pregnancy, the PES and WDEQ‑A questionnaires, as well as birth preference form were further completed. Mode of birth was recorded in the first few days of postpartum. The Fisher's exact test, along with ANOVA and Chi-square tests were used to determine associations between variables. A paired t-test was used to examine within-group comparisons. The Kruskal-Wallis non-parametric test was used to investigate the intervening effect of economic status.ResultsPost intervention, the mean score of pregnancy experience and FOC did not differ significantly between the three groups. Also, 86.8% of participants in group A, 90.4% of participants in group B, and 62% of participants in the control group preferred to give birth vaginally, which was statistically significant (p = 0.001). Moreover, 66% of participants in group A, 61.5% of participants in group B, and 50% of participants in the control group ultimately gave birth vaginally. None of the participants in group A underwent an elective cesarean section, while this rate was 7.7% and 24% for groups B and control, respectively (p = 0.002).ConclusionDespite the non-significant differences identified between the three groups in terms of pregnancy experience and FOC, prenatal education delivered via social media may be usefully offered to Iranian women keen to receive education flexibly online.Trial registrationName of the Registry: Iranian Registry of Clinical Trials. Trial registration number: IRCT20180427039436N2. Date of registration: 15/06/2018. URL of trial registry record: https://www.irct.ir/trial/30890.

Dataset Information

AI and narrative embeddings detect PTSD following childbirth via birth stories.

Publications

AI and narrative embeddings detect PTSD following childbirth via birth stories.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets