What question did this study set out to answer?

The study aims to predict mortality and injury rates resulting from attacks on healthcare facilities in Syria.

March 25, 2026Prehospital and Disaster Medicine0 citations

Predicting Mortality and Injury Rates in Healthcare Facility Attacks in Syria: A Bayesian Ridge Regression Approach

HSHeejun Shin RBRadwan BarbandiSoonchunhyang University

Key Points

The study aims to predict mortality and injury rates resulting from attacks on healthcare facilities in Syria.
Utilized a Bayesian Ridge Regression model to analyze data from the White Helmets on attacks.
Encoded categorical variables and extracted temporal variables for analysis.
Evaluated the influence of geographic, demographic, and temporal factors on outcomes.
The model indicated that deceased men, women, and children significantly affect mortality prediction.
Homs, Idleb, and Hama governorates showed substantial impacts on both mortality and injury rates.
Achieved R-squared values of 1.0 for mortality and 0.98 for injury rate predictions, indicating strong accuracy.

Abstract

Introduction: Attacks on hospitals and medical points in conflict zones significantly impact mortality and injury rates. This study applies a Bayesian Ridge Regression model to predict these rates, analyzing variables such as geographic location, type of operation, perpetrator, and demographics. Methods: The dataset, collected by the White Helmets, comprised records of attacks, detailing injuries, fatalities, and contextual variables. Categorical variables (Governorate, Type of Operation, Perpetrator) were encoded, and temporal variables (Year, Month, Day) were extracted. The Bayesian Ridge Regression model was developed, trained, and tested to determine the influence of each factor on mortality and injury rates. Results: For mortality prediction, the model identified the number of deceased men (β = 1.23), women (β = 0.57), and children (β = 0.40) as the most influential variables. Geographic locations, particularly Homs (β = 1.23), Idleb (β = 0.57), and Hama (β = 0.40) governorates, also had significant impacts. Temporal factors showed minimal influence, with coefficients for the year (β = 4.47e-09), month (β = 4.69e-09), and day (β = 2.50e-08). The model demonstrated high accuracy with an R-squared value of 1.0. For injury rate prediction, the number of injured men (β = 1.64e-10), women (β = -1.23e-10), and children (β = 1.18e-10) were influential. Homs (β = 1.23), Idleb (β = 0.57), and Hama (β = 0.40) also significantly impacted injury rates. Temporal factors had low coefficients, with year (β = 3.24e-10), month (β = 3.17e-10), and day (β = 1.15e-09). The model achieved an R-squared value of 0.98, indicating strong predictive capability for injury rates. Conclusion: The Bayesian Ridge Regression model provides critical insights into factors affecting mortality and injury rates in attacks on healthcare facilities. These findings can inform clinical guidelines, emphasizing the protection of vulnerable populations, resource allocation in high-risk regions, and implementing targeted interventions based on these results.

AIに質問

Bookmark

View Full Paper