Neural Network Computer Analysis of Fetal Heart Rate

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Obstetrics and Gynecology

Background:

• Intrapartum fetal heart rate (FHR) monitoring is crucial for assessing fetal well-being during labor.
• Objective evaluation of FHR patterns remains a challenge, often relying on subjective interpretation.
• Developing automated, objective FHR analysis tools is essential for improving clinical decision-making.

Purpose of the Study:

• To evaluate the nonsubjective performance of a neural network (NNW) computer system for intrapartum FHR analysis.
• To assess the clinical applicability of the NNW system in predicting fetal outcomes.
• To compare the NNW system's accuracy with conventional FHR analysis methods.

Main Methods:

• A neural network (NNW) computer system was developed with input, intermediate, and output layers for FHR data processing.
• The NNW system underwent extensive training using 50-minute FHR data from 20 cases with known outcomes.
• The trained NNW system was tested on 29 new FHR cases, with outcome probabilities displayed as trendgrams.

Main Results:

• The NNW system achieved 100% accuracy in internal validation and 86% correct evaluation in external validation against cardiotocogram, Apgar score, and umbilical arterial pH.
• Trendgrams generated by the NNW system accurately reflected typically normal or abnormal cases, with intermediate probabilities indicating transitional states.
• Mixed suspicious and pathological probabilities from the NNW system correlated with pathological fetal outcomes.

Conclusions:

• The NNW FHR analysis system demonstrates promising capabilities for objective decision-making during the intrapartum period.
• Outcome probabilities and trendgrams provided by the NNW system offer valuable insights into fetal status.
• This AI-driven approach has the potential to enhance the accuracy and objectivity of intrapartum fetal monitoring.