Purwanto Purwanto, Ike Sari Astuti, Ardyanto Tanjung, Fatchur Rohman, Kresno Sastro Bangun Utomo
This study aimed at developing a machine learning-based COVID-19 transmission risk spatial model and an analysis of the public spatial awareness in the use of personal protective equipment (PPE) in their spatial environment (i.e., a spatial-based COVID-19 transmission risk model). Random Forest model combined with Information Gain was used in this study. Twenty-three geospatial variables that passed a feature selection process were inputted to build a spatial-based COVID-19 transmission risk model. The validation outcome reveals that Random Forest (RF) model achieved excellent COVID-19 transmission risk prediction result (AUC value of 94.7%). Transmission risk modeling results show that the moderate-to-high class risk pattern was concentrated in residential areas with complex road infrastructure. The model showed that land use, road infrastructure, minimum temperature, and close to urban areas is important for spatial modeling of COVID-19 transmission risk. The results of the public spatial awareness analysis show variations in respondents' use of PPE based on their spatial environment. They tend to carry one PPE whose use increases significantly from low-risk to high-risk. While being indoor, the majority used hand sanitizer, and a face mask when being outdoor. Thus, developing a spatial-based COVID-19 transmission risk prediction linked with public spatial awareness can be used as a reference to establish health policies or regulations by decision-makers, health boards, and governments to control and simultaneously reduce the risk of transmission more effectively and efficiently. © 2023 Elsevier Ltd
Department of Geography, Faculty of Social Sciences, Universitas Negeri Malang, Jl. Semarang 5, Malang, 65145, Indonesia; Department of Biology, Faculty of Mathematics and Natural Sciences, Universitas Negeri Malang, Jl. Semarang 5, Malang, 65145, Indonesia