Year: 2025
Type: article
Study design: Observational Study
Abstract: Objective: Our research aims to propose the application of AI in improving the accuracy for estimating fetal weight. Method: This is a cross-sectional analytical study combining the development and comparison of predictive models supported by AI that include 1162 pregnancy women that came for scheduled visit and give birth at Obstetric Department at Vinmec Times City International Hospital. The pregnancy women were chosen based on: age (18 to 35 years old), live singleton pregnancy, Vietnamese, height > 153 cm, and gestational age ≥ 36 weeks. Maternal and fetal parameters are then collected and divided into data and test set. Symbolic regression learning method is applied to our ensemble model which include Operon, XGBoost, LightGBM, FEAT, Hadlock, GP-GOMEA, DSR, and SBP-GP. The performance of each algorithms are then evaluated using test data set and compared the results to the traditional method. Results: 1162 pregnancy women were selected, and each were collected 9 maternal and fetal variables. MAE, MSE, RMSE, and R2 value of each model were calculated from their performance. Operon model’s MAE was 158.542 grams, RMSE was 205.485 and R2 of 0.538, while Hadlock model’s MAE was 186.98 grams, RMSE was 331.373 and R2 of -0.19. Conclusion: AI-driven methods, particularly symbolic regression, demonstrates greater accuracy and efficacy compared to the traditional method, hence suggests a promising potential in prenatal weight estimation.
Source: Tạp chí Y học Việt Nam
Authors: Ba Nha Pham, Phuong Tu Nguyen, Le Duc Anh Tran, Ba Quang Huy Do, Hai Phuong Nguyen, Anh Thu Nguyen, Phuong Hong Minh Nguyen, Thien Thai Le, Thi Huong Linh Nguyen, Tien Hoang Nguyen
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W4409932180