Large language model-assisted machine learning for postpartum urinary incontinence prediction: Model development and validation study - Summary - MDSpire
To develop and validate a postpartum urinary incontinence prediction model combining large language model (LLM)-assisted extraction of electronic medical record information with machine learning algorithms.
Approach:
Design and participants: Prospective investigation using development data from 1,243 eligible women consecutively recruited at one campus of a tertiary maternity hospital in Hangzhou from January through June 2020, and external validation data from 452 eligible women recruited at a second campus from January through May 2025.
Outcome: Urinary incontinence at 6–8 weeks postpartum was assessed using the validated Chinese ICIQ-UI SF. Participants reporting any urinary leakage were categorized as having postpartum UI.
Key Findings:
The supplied excerpt describes the study rationale, participant cohorts, outcome definition, and methods available before the text truncates; it does not report model performance or study results.
Interpretation:
The study aimed to develop and validate a postpartum urinary incontinence prediction model integrating LLM-assisted extraction of EMR information with machine learning algorithms.
Limitations:
The provided context ends during the methods section and does not include model-development details, validation results, or the authors’ stated limitations.
