Advancing Pediatric Growth Assessment with Machine Learning: Overcoming Challenges in Early Diagnosis and Monitoring

Area of Science:

• Pediatric Healthcare
• Medical Informatics
• Machine Learning Applications

Background:

• Accurate pediatric growth assessment is vital for early detection and management of growth disorders.
• Traditional methods for growth assessment often lack precision and real-time capabilities.
• Machine learning (ML), particularly logistic regression, presents an opportunity to enhance diagnostic accuracy and timeliness.

Purpose of the Study:

• To evaluate the efficacy of a logistic regression model in improving pediatric growth assessment.
• To demonstrate the potential of interpretable ML models in clinical practice.
• To compare ML-based assessment with traditional methods.

Main Methods:

• Development of a logistic regression model using R on a cross-sectional dataset.
• Application of data preprocessing techniques including cleaning, imputation, and feature selection.
• Evaluation of model performance using metrics such as accuracy, sensitivity, and ROC curve analysis.

Main Results:

• The logistic regression model achieved 94.65% accuracy and 91.03% sensitivity in identifying growth anomalies.
• The model's ROC curve yielded an Area Under the Curve (AUC) of 0.96, indicating excellent predictive power.
• ML demonstrates potential for automating pediatric growth monitoring and supporting clinical decisions due to its simplicity and interpretability.

Conclusions:

• Logistic regression offers a promising ML tool for enhancing diagnostic precision and operational efficiency in pediatric healthcare.
• Challenges include data quality, clinical integration, and privacy concerns.
• Future research should address dataset diversity, model interpretability, and external validation for wider clinical adoption.