Research on Multi-Feature Intelligent Prenatal Fetal Monitoring Algorithm Based on the PSO-RF Model

In prenatal fetal state monitoring, fetal heart rate and uterine contraction data serve as important assessment criteria, yet they face data classification challenges. This paper proposes the PSO-RF algorithm, aiming to improve the prediction accuracy of prenatal fetal monitoring. By using Particle Swarm Optimization (PSO) to automatically search for the optimal hyperparameter combination, the complexity of manual hyperparameter tuning is avoided. Meanwhile, the SMOTE oversampling technique is applied to balance the dataset, reducing the impact of data imbalance on model performance. The Isolation Forest algorithm is used to remove anomalous samples, improving data quality. The results indicate that compared to traditional models, the PSO-RF model shows a 2.2% improvement in precision, a 0.2% increase in accuracy, a 1.4% improvement in recall, and a 0.9% increase in the F1 score, especially demonstrating higher classification accuracy and lower computational overhead on imbalanced datasets. This method effectively enhances the performance of Random Forest (RF) on complex datasets. However, there is still room for further optimization in feature selection and applications involving large-scale datasets. Future research could combine other optimization methods and deep learning techniques to further enhance model accuracy and efficiency, and explore its potential applications in other fields.