A joint design for functional data with application to scheduling ultrasound scans

Area of Science:

• Biostatistics
• Maternal-Fetal Medicine
• Functional Data Analysis

Background:

• Accurate fetal growth assessment is crucial for predicting child birth outcomes.
• Current methods for collecting functional data, like ultrasound measurements, may not be optimized for joint prediction tasks.
• There is a need for a design that simultaneously considers the recovery of growth trajectories and prediction of scalar outcomes.

Purpose of the Study:

• To propose a joint design for sampling functional data that optimizes the prediction of both the functional data and a scalar outcome.
• To apply this joint design to the specific problem of fetal growth monitoring using ultrasound data.
• To determine optimal timing for ultrasound measurements to enhance fetal growth trajectory recovery and predict birth outcomes.

Main Methods:

• Formulation of a joint design using an optimization criterion.
• Implementation of the proposed design in a pilot study.
• Evaluation of the design's performance through simulation studies and real-world application to fetal ultrasound data.

Main Results:

• The proposed joint design demonstrates effectiveness in achieving optimal prediction for both functional data and scalar outcomes.
• The pilot study and simulations confirm the practical utility and performance of the joint sampling design.
• The method provides insights into optimal timing for data collection in longitudinal studies.

Conclusions:

• A novel joint design for functional data sampling is presented, offering improved predictive accuracy.
• The approach is particularly valuable for applications like fetal growth monitoring, enhancing both trajectory estimation and outcome prediction.
• This methodology can guide efficient data collection strategies in complex biomedical studies.