Detection of Monkeypox Cases Based on Symptoms Using XGBoost and Shapley Additive Explanations Methods

Area of Science:

• Computational epidemiology
• Machine learning in public health
• Infectious disease diagnostics

Background:

• The monkeypox virus presents a significant global public health threat with pandemic potential.
• Machine learning (ML) has demonstrated efficacy in diagnosing various diseases, including cancer and COVID-19.
• Existing monkeypox diagnostic research primarily focuses on image-based analysis.

Purpose of the Study:

• To develop and evaluate ML models for monkeypox diagnosis based on textual symptom data.
• To compare the performance of multiple ML algorithms for this specific diagnostic task.
• To provide an interpretable ML model for monkeypox diagnosis.

Main Methods:

• A textual dataset was curated using data from Global Health and the World Health Organization (WHO).
• Several ML algorithms were employed, including Extreme Gradient Boosting (XGBoost), CatBoost, LightGBM, Support Vector Machine (SVM), and Random Forest.
• K-fold cross-validation and Shapley Additive Explanations (SHAP) were utilized for model evaluation and interpretability.

Main Results:

• Extreme Gradient Boosting (XGBoost) demonstrated superior performance, achieving an accuracy of 1.0.
• K-fold cross-validation confirmed the model's robustness, yielding an average accuracy of 0.9 across 5 splits.
• SHAP analysis provided insights into the XGBoost model's decision-making process.

Conclusions:

• ML models, particularly XGBoost, can effectively diagnose monkeypox using symptom-based textual data.
• This approach offers a novel alternative to image-based diagnostic methods.
• The developed model shows promise for rapid and accurate public health surveillance of monkeypox.