Data Driven National Decision Support System for Management of Communicable Diseases

Abstract

Infectious disease syndrome like COVID-19 falls under the Public health domain and needs to be addressed with timely decisions and rapid actions. For such diseases, the dispersal becomes exponential with frequent social gatherings, therefore the immediate strategy, to control the surging waves of COVID-19, comes with perception to impose immediate lockdown in infected zones. Preceding pandemic trends uncovered lethal pathogenic (like Corona Virus) infections that created havoc on existing healthcare structures by infecting humans in large numbers thereby overwhelming existing healthcare management models and warning us of similar medical escalations inevitable in the near future as well. As a result, an urgent need was realized to build a resilient pandemic-lockdown model to relieve health systems from crashing and enable them to be equipped with the latest tools and techniques for rapid actions to manage future pandemics efficiently. The concept of street networks incorporated with shortest path algorithm e.g. minimum spanning tree (MST) provides a broader vision to define an approach to investigate the correlation between reported infectious cases at street segment level in order to highlight the exact areas of concern. Therefore, this combination helps the authorities to adopt better strategy to lockdown the targeted areas in unplanned colonies. Geo-spatial representation comes with subsequent composition of patterns to identify the particular streets in more granular means of visualization that aids in the precise lockdown. Therefore, this research work aims at proposing an adaptable framework called the automated selection of streets for lockdown (ASSLD) to mitigate the exponential progression of the epidemic from the perspective of urban resilience. Intuitive selection of pandemic-affected streets was accomplished by first applying the shortest path algorithm over the respective spatial network and then validating the selection through spatial analysis using space syntax syntactic measures Medical records for COVID-19 patients of a preceding contagious viral infection were used as a dataset to prepare test beds against which the ASSLD/ ALDSS model was regressively validated. Multiple linear regression was used to analyse the impact of space syntax syntactic measures along with socio-economic factors on the number of replicating COVID-19 patients. The findings indicated that space syntax syntactic variables along with certain socioeconomic factors, provided more granular means of measuring the spatial elements (streets) as diffusion hubs for a communicable disease. This thesis presents in detail proof of concept with results applied to actual data obtained from a preceding communicable syndrome outbreak. In this thesis, the concept of street networks is incorporated with shortest path algorithm e.g. minimum spanning tree (MST) to define an approach to investigate the correlation between reported COVID-19 cases and relevant streets in order to adopt better lockdown strategy for unplanned colonies. Geo-spatial representation has been used for subsequent composition of patterns to identify the particular streets for locked down. Results show that MST provides better solution by evaluating explicit areas of concern for lockdown plans.