Feature selection for computerized fetal heart rate analysis using genetic algorithms.

Area of Science:

• Medical Informatics
• Computational Biology
• Obstetrics

Background:

• Timely diagnosis during birth is crucial for preventing conditions like birth asphyxia.
• Fetal heart rate (FHR) monitoring is standard for assessing fetal well-being during labor.
• Computerized FHR analysis can assist clinicians in identifying abnormal patterns and guiding interventions.

Purpose of the Study:

• To apply Genetic Algorithms (GA) for feature selection in FHR analysis.
• To identify an optimal subset of FHR features for recognizing unfavorable patterns.
• To integrate selected features for improved classification of FHR data.

Main Methods:

• Utilized Genetic Algorithms (GA) to select the best feature subset from 64 FHR features.
• Trained the GA on 408 balanced cases and tested on 102 balanced cases.
• Employed a linear Support Vector Machine (SVM) classifier and formed a committee of 100 classifiers.

Main Results:

• Achieved fair classification performance on the testing set.
• Reported a Cohen's kappa of 0.47 and a proportion of agreement of 73.58%.
• Demonstrated the application of feature selection for FHR analysis on a large database.

Conclusions:

• Genetic Algorithms (GA) can effectively select relevant features for FHR pattern recognition.
• This approach shows potential for enhancing the accuracy of computerized FHR analysis.
• This study represents a novel application of feature selection methods in FHR analysis using a substantial dataset.