Identifying key soil characteristics for Francisella tularensis classification with optimized Machine learning models

Area of Science:

• Environmental microbiology
• Computational epidemiology
• Bioinformatics

Background:

• Francisella tularensis (Ft) is a significant bioweapon threat impacting animal and human health.
• Traditional statistical methods limit understanding of Ft's soil-based epidemiology.
• Advanced computational approaches are needed for accurate pathogen classification and environmental risk assessment.

Purpose of the Study:

• To develop and optimize machine learning (ML) models for predicting epidemiological models of soil-borne Francisella tularensis.
• To identify crucial soil physical-chemical attributes influencing Ft presence and persistence.
• To enhance pathogen classification accuracy and reduce false positives.

Main Methods:

• A two-stage feature ranking process was implemented to identify significant soil attributes.
• Hyperparameter optimization was performed using Bayesian and Random search techniques.
• Classification algorithms including Support Vector Machines (SVM), Ensemble Models (EM), and Neural Networks (NN) were evaluated.

Main Results:

• Key soil features influencing Ft classification include clay, nitrogen, soluble salts, silt, organic matter, and zinc.
• Bayesian optimization achieved the highest accuracy: 86.5% for SVM, 81.8% for EM, and 83.8% for NN.
• Support Vector Machines (SVM) demonstrated superior performance, reaching 86.5% accuracy with both optimization methods.

Conclusions:

• Machine learning models significantly enhance the prediction of soil-based microbial epidemiological models.
• Identifying critical soil factors improves understanding of Ft environmental persistence and reduces misclassification risks.
• This research contributes to improved pandemic preparedness and mitigation of socio-economic impacts.