An App Detecting Dengue Fever in Children: Using Sequencing Symptom Patterns for a Web-Based Assessment

Area of Science:

• Medical Informatics
• Epidemiology
• Pediatrics

Background:

• Dengue fever (DF) is a widespread arthropod-borne viral illness.
• Current DF diagnosis relies on basic symptom presence, lacking predictive nuance.
• An accessible tool is needed for early DF identification.

Purpose of the Study:

• To develop a mobile application for predicting Dengue Fever risk.
• To identify key clinical symptoms for early DF detection.

Main Methods:

• Analysis of 17 DF-related symptoms in 177 pediatric patients.
• Utilized unweighted summation score, nonparametric HT statistic, and logistic regression.
• Extracted statistically significant predictors for DF risk.

Main Results:

• Six symptoms (family history, high fever, rash, petechiae, abdominal pain, weakness) significantly predicted DF.
• A combined model achieved 0.87 sensitivity and 0.84 specificity (AUC 0.91).
• Predictive specificity improved by 10.2% compared to traditional logistic regression.

Conclusions:

• Six simple symptoms effectively predict pediatric Dengue Fever risk.
• Combining weighted regression and HT coefficient enhances predictive accuracy.
• A mobile app can facilitate early DF self-assessment, potentially reducing reliance on lab tests.