Severe Dengue Prognosis Using Human Genome Data and Machine Learning.

Area of Science:

• Genetics
• Immunology
• Computational Biology

Background:

• Dengue is a major global arthropod-borne disease.
• Current diagnostic methods for dengue phenotypes are often inaccurate.

Purpose of the Study:

• To develop a machine learning model for predicting dengue fever severity using only human genome data.
• To identify genetic markers associated with dengue severity.

Main Methods:

• Genotyping of 102 Brazilian dengue patients and controls for 322 innate immunity single nucleotide polymorphisms (SNPs).
• Utilizing a support vector machine (SVM) for feature selection and an artificial neural network (ANN) for classification.
• Training the ANN on 13 key immune SNPs selected under dominant or recessive models.

Main Results:

• The ANN model achieved median accuracy greater than 86%.
• The model demonstrated high sensitivity (over 98%) and specificity (over 51%) in classifying dengue fever versus severe dengue.
• Identified 13 key immune SNPs crucial for accurate classification.

Conclusions:

• Human genome data alone can accurately predict dengue fever severity.
• This genetic-based classification method can identify individuals at high risk for severe dengue, even before infection.
• Genetic factors play a significant role in defining dengue phenotypes and can be leveraged for disease prediction and management.