Detecting dengue fever in children using online Rasch analysis to develop algorithms for parents: An APP development...

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Public Health Surveillance

Background:

• Dengue fever (DF) poses a significant public health challenge in Asia.
• Traditional DF detection methods are often difficult.
• Machine learning models like CNNs and ANNs show promise for improving prediction accuracy.

Purpose of the Study:

• To investigate the effectiveness of combining CNN, ANN, KNN, and LR for DF prediction in children.
• To explore item features and responses using online Rasch analysis for DF detection.
• To develop a practical tool for early DF diagnosis in pediatric patients.

Main Methods:

• Extracted 19 DF symptom variables from 177 pediatric patients (69 with DF).
• Utilized Rasch analysis to examine 11 variables for DF risk prediction.
• Calculated prediction accuracy (AUC) using training (80%) and testing (20%) datasets, comparing combined algorithms vs. individual ones.

Main Results:

• Rasch analysis provided easily interpretable visual displays of DF data.
• Logistic Regression (LR) achieved a relatively high AUC (0.70).
• An app was developed to assist parents in detecting DF in children.

Conclusions:

• An LR-based mobile application for pediatric DF detection has been successfully developed.
• An 11-item model is proposed to aid early differentiation of DF from other febrile illnesses.
• The app aims to support patients, families, and clinicians in timely DF diagnosis.