Scientific Reports · 8 authors, 8 centres
This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.
This retrospective study developed support vector regression and random forest models to predict fetal RR intervals from 13 maternal physiological features. Support vector regression outperformed random forest, yielding significant correlations between predicted and actual fetal RR intervals with error percentages generally at or below 5%.
The study used a retrospective design to develop two machine learning models, support vector regression (SVR) and random forest (RF), for predicting fetal RR intervals (fRRI) from 13 maternal features including age, weight, and electrocardiogram-derived parameters. Pregnant women with maternal complications were included. SVR outperformed RF, with 14 test subjects showing error percentages at or below 5% in most cases, though three subjects exceeded this threshold. Correlations between predicted and actual fRRI were significant but had coefficient values lower than 0.5. Shapley analysis identified maternal age, RRI, SDHR, and HF as the features with highest impact on prediction. Limitations include the retrospective design, maternal complications that may have affected accuracy, and exclusion of available features such as blood pressure and respiratory rate. The authors suggest that integrating maternal conditions could enhance fetal health assessment.
