Prediction of Shigellosis outcomes in Israel using machine learning classifiers

Area of Science:

• Epidemiology
• Infectious Diseases
• Machine Learning in Healthcare

Background:

• Shigellosis is a major cause of childhood morbidity and mortality globally.
• Trends in Shigellosis incidence and species distribution in Israel (2002-2015) were previously unevaluated.
• Disparities in Shigella infection rates were observed across different ethnic groups in Israel.

Purpose of the Study:

• To analyze Shigellosis epidemiological trends in Israel from 2002 to 2015.
• To identify demographic and clinical factors associated with Shigella infection and hospitalization.
• To develop machine learning models for predicting Shigella species and hospitalization outcomes.

Main Methods:

• Analysis of national notifiable disease reporting data from 2002-2015.
• Statistical comparison of Shigellosis incidence across age groups and ethnic populations.
• Development and validation of machine learning algorithms using demographic and clinical data.

Main Results:

• Shigella sonnei was the predominant species; Shigella flexneri showed higher hospitalization rates.
• Infants and young children (0-5 years) exhibited the highest Shigella morbidity.
• Significantly higher Shigella flexneri incidence was noted among Muslims compared to other ethnic groups.

Conclusions:

• Machine learning models achieved high accuracy (93.2% for species, 94.9% for hospitalization).
• These algorithms can potentially enhance clinical diagnosis and treatment strategies for Shigellosis.
• Understanding epidemiological trends and ethnic disparities is crucial for public health interventions.