Fetal weight estimation for prediction of fetal macrosomia: does additional clinical and demographic data using...

Area of Science:

• Medical imaging
• Machine learning in healthcare
• Obstetrics and Gynecology

Background:

• Accurate prediction of fetal macrosomia is crucial for optimizing pregnancy outcomes.
• Current sonographic estimation of fetal weight relies on regression-based formulas with limitations in predictive accuracy.

Purpose of the Study:

• To evaluate if pattern recognition classifiers using clinical and sonographic data improve ultrasound prediction of fetal macrosomia compared to traditional formulas.
• To assess the performance of a Support Vector Machine (SVM) algorithm in predicting fetal macrosomia.

Main Methods:

• The study employed the SVM algorithm for binary classification of fetal weight estimation (>4000g vs. <4000g).
• Clinical and sonographic variables from 100 pregnancies suspected of having large for gestational age (LGA) fetuses were analyzed.
• Feature selection identified 13 significant variables out of 38 for distinguishing birth weights.

Main Results:

• The SVM algorithm achieved a sensitivity of 81%, specificity of 73%, positive predictive value of 81%, and negative predictive value of 73% for predicting macrosomia (cutoff 4000g).
• Comparative analysis using two common regression-based formulas yielded 88.1% sensitivity, 34% specificity, 65.8% positive predictive value, and 66.7% negative predictive value.

Conclusions:

• The SVM algorithm demonstrates comparable prediction of LGA fetuses to established regression-based formulas.
• The enhanced specificity and positive predictive value of the SVM method suggest its potential clinical utility.
• Further data accumulation is recommended to improve the reliability and validation of this machine learning approach for fetal weight prediction.