A systematic review of artificial intelligence frameworks for holistic disaster management

Abstract This work gives a concise literature review of the previous studies on disaster management through the use of Artificial Intelligence (AI). It is supported by evidence from 96 peer-reviewed articles sourced from multiple academic databases, including IEEE Xplore, Elsevier, Wiley and Springer. It discusses two broad topics: The phases of the disaster management cycle (Preparedness, Response, and Recovery) and a three-stage time frame (Primitive: 2000–2010, Transitional: 2011–2016, Modern: 2017-present). The research is conducted through visual graphics to demonstrate variation in research activity in different regions, periods, and phases of disasters. It involves a bibliometric co-occurrence analysis using VOSviewer, aimed at obtaining important thematic clusters. Tables focus on comparing the performance of various AI methods applied to particular disaster-related tasks. The findings indicate the obvious replacement of conventional rule-based solutions with more sophisticated deep learning solutions. It throws light on existing gaps and recent developments by describing methodological trends, outcomes of performance and future directions of research.