Knowledge and Information Systems Human-Computer Interaction

Rapid evaluation of infection probability in ventilation systems based on heuristic algorithms and a constrained knapsack model

Hsu, Hsieh-Chih, Lee, Hsin-Chun, Pan, Chen-Yu, Cho, Yen-Cheng, Huang, Ko-Wei
Springer Nature · 2026-11-18 · DOI 10.1007/s10115-026-02937-5

Abstract

Rapid identification of high-risk spaces in existing building ventilation systems remains challenging due to insufficient design information and the high computational cost of numerical simulations. This study proposes a rapid method for assessing pollutant dispersion and infection probability by transforming ventilation system parameters into a constrained knapsack model and solving it directly using heuristic algorithms, namely, the Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). The results indicate that both algorithms exhibit a high degree of consistency in the overall infection risk distribution trends, which are also aligned with those reflected by the Wells–Riley model. Rather than replacing computational fluid dynamics (CFD) simulations, which provide detailed airflow information, the proposed approach serves as a computationally efficient screening tool for identifying high-risk rooms prior to detailed analysis.

Read at publisher →