Functional data analysis with application to periodically stimulated foetal heart rate data. I: functional regression

Area of Science:

• Biostatistics
• Developmental Psychology
• Perinatal Medicine

Background:

• Longitudinal data analysis often faces challenges with a high number of measurements per subject.
• Functional regression is a statistical technique suited for such data, treating it as singular longitudinal analysis.
• Traditional analysis of fetal heart rate data relies on subjective concepts like habituation.

Purpose of the Study:

• To modify existing functional regression techniques for a functional covariate with repeated stimuli.
• To apply this modified approach to analyze periodically stimulated fetal heart rates.
• To predict child psychomotor development at 18 months of age using fetal heart rate tracings.

Main Methods:

• Development of a modified functional regression model.
• Application of the model to fetal heart rate tracings from periodically stimulated fetuses.
• Using complete heart rate tracings as predictors for psychomotor development.

Main Results:

• The modified functional regression successfully utilized entire fetal heart rate tracings.
• This approach eliminated the need for a subjective definition of habituation.
• All available information from heart rate data was leveraged for prediction.

Conclusions:

• Modified functional regression offers a robust method for analyzing complex longitudinal data like fetal heart rates.
• This technique enhances the prediction of child psychomotor development by fully utilizing predictor information.
• The study provides a more objective and comprehensive approach compared to traditional habituation-based analyses.