Hybrid Machine Learning Models for Predicting Types of Human T-cell Lymphotropic Virus.

Area of Science:

• Bio-informatics and Computational Biology
• Virology and Infectious Diseases
• Machine Learning Applications in Healthcare

Background:

• Human T-cell Lymphotropic virus (HTLV) causes severe diseases including adult T-cell leukemia and HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP).
• HTLV-1 affects over 20 million people globally, with limited diagnostic methods and no current vaccine, leading to many asymptomatic carriers.
• Accurate and timely detection of HTLV types is crucial for disease management and epidemiological control.

Purpose of the Study:

• To develop novel, efficient, and rapid diagnostic methods for Human T-cell Lymphotropic virus (HTLV) detection.
• To propose and evaluate 64 hybrid machine learning techniques for predicting HTLV types (HTLV-1, HTLV-2, HTLV-3).
• To enhance the current diagnostic workflow for HTLV infections.

Main Methods:

• Development of 64 hybrid machine learning models by combining four classification, four feature weighting, and four feature selection techniques.
• Evaluation of model performance using standard metrics such as accuracy, AUROC, and F1 score.
• Application of models to predict HTLV types from biological data, potentially protein sequences.

Main Results:

• The proposed hybrid machine learning models demonstrated high efficiency in predicting HTLV types.
• The best hybrid model achieved an accuracy of 99.85%, an AUROC of 0.99, and an F1 score of 0.99.
• These models show potential for supporting and verifying results from complex confirmatory tests like Western blotting.

Conclusions:

• Hybrid machine learning techniques offer a powerful and accurate approach for HTLV type prediction.
• The developed models can significantly assist in the timely and efficient diagnosis of HTLV-1 infections.
• Further exploration of physicochemical properties of HTLV protein sequences using these models can yield deeper insights into the virus.