Comparison of new modeling methods for postnatal weight in ELBW infants using prenatal and postnatal data
Peter J Porcelli1, S Trent Rosenbloom
1*Department of Pediatrics, Wake Forest University, Winston-Salem, NC †Department of Biomedical Informatics, Vanderbilt University, Nashville, TN.
Insights
Artificial neural networks (NN) create more accurate postnatal infant weight curves than traditional methods. This improves assessment of growth and fluid management for extremely-low-birth-weight infants.
Area of Science:
- Neonatalogy
- Medical Informatics
- Biostatistics
Background:
- Postnatal infant weight curves are crucial for assessing fluid management, nutrition, and growth.
- Traditional weight curves rely solely on birth weight, neglecting vital postnatal clinical data.
- Accurate weight assessment is particularly critical for extremely-low-birth-weight (ELBW) infants.
Purpose of the Study:
- To compare the accuracy of traditional birth weight-based infant weight curves with curves generated from individual patient records.
- To evaluate two predictive modeling methods: linear regression (LR) and artificial neural networks (NN).
- To determine if electronic health record data can improve postnatal weight curve accuracy.
Main Methods:
- Collected perinatal demographic and postnatal nutrition data for 92 ELBW infants (birth weight <1000 g).
- Generated static weight curves using published algorithms.
- Developed predictive models using linear regression and artificial neural networks with patient-specific data.
Main Results:
- Artificial neural network (NN) models demonstrated significantly higher accuracy in predicting infant weight compared to static curves and linear regression (LR).
- The NN model yielded a mean absolute residual of 12.9 ± 9.2 g, substantially lower than LR (60.9 ± 49.1 g) and static curves (84.8 ± 74.4 g).
- Infants receiving nothing by mouth (NPO) exhibited greater discrepancies in weight curve predictions.
Conclusions:
- Artificial neural network (NN)-generated weight curves more accurately reflect the actual weight status of extremely-low-birth-weight infants.
- Utilizing electronic health record data with NN models offers a superior approach to creating postnatal weight curves.
- This advanced method enhances the assessment of growth and fluid management in vulnerable ELBW populations.
Objectives:
Postnatal infant weight curves are used to assess fluid management and evaluate postnatal nutrition and growth. Traditionally, postnatal weight curves are based on birth weight and do not incorporate postnatal clinical information. The aim of the present study was to compare the accuracy of birth weight-based weight curves with weight curves created from individual patient records, including electronic records, using 2 predictive modeling methods, linear regression (LR) and an artificial neural network (NN), which apply mathematical relations between predictor and outcome variables.
Methods:
Perinatal demographic and postnatal nutrition data were collected for extremely-low-birth-weight (ELBW; birth weight <1000 g) infants. Static weight curves were generated using published algorithms. The postnatal predictive models were created using the demographic and nutrition dataset.
Results:
Birth weight (861 ± 83 g, mean ± 1 standard deviation [SD]), gestational age (26.2 ± 1.4 weeks), and the first month of nutrition data were collected from individual health records for 92 ELBW infants. The absolute residual (|measured-predicted|) for weight was 84.8 ± 74.4 g for the static weight curves, 60.9 ± 49.1 g for the LR model, and 12.9 ± 9.2 g for the NN model, analysis of variance: both LR and NN P<0.01 versus static curve. NPO (nothing by mouth) infants had greater weight curve discrepancies.
Conclusions:
Compared with birth weight-based and logistic regression-generated weight curves, NN-generated weight curves more closely approximated ELBW infant weight curves, and, using the present electronic health record systems, may produce weight curves better reflective of the patient's status.
