Classification of dengue illness based on readily available laboratory data

Area of Science:

• Medical Diagnostics
• Infectious Diseases
• Clinical Laboratory Science

Background:

• Dengue fever diagnosis often relies on clinical signs, imaging, and hemoconcentration.
• Accurate classification of dengue illness is crucial for timely and appropriate patient management.

Purpose of the Study:

• To evaluate the effectiveness of clinical laboratory data for retrospective dengue illness classification.
• To develop and validate predictive models for distinguishing dengue hemorrhagic fever (DHF) from dengue fever (DF) and other febrile illnesses (OFI).

Main Methods:

• Analysis of clinical laboratory data from 1,227 children with acute febrile illness in Thailand.
• Utilized multivariable logistic regression to build classification models.
• Validated models using data from a separate hospital cohort.

Main Results:

• Models achieved high sensitivity, ranging from 89.2% (dengue vs. OFI) to 79.6% (DHF vs. DF).
• Classification models demonstrated strong performance on the validation dataset.
• Developed probability-based classification using accessible laboratory markers.

Conclusions:

• Clinical laboratory data alone can effectively classify dengue illness retrospectively.
• These models offer a promising, accessible tool for dengue diagnosis in resource-limited settings.
• Further validation in diverse dengue-endemic regions is recommended.