Developing predictive technologies for the strategic management of structural fire disasters

House (structure) fires pose a recurring threat to life and property in every society, such that common traditions of fire-fighting services and fire Emergency Management developed over many centuries. From such traditions, there has evolved the Comprehensive Emergency Management Framework (National Governors Association, 1979) encompassing Prevention, Preparation, Response, Recovery, (PPRR) whose activities are continually enhanced through improvements in information systems. Such systems now incorporate a repository of large datasets gathered over years of house (structure) fire incidents. Motivated by the devastation of recent major fire emergencies in their jurisdiction, the named M2inder project was initiated by a state-based emergency service organization to develop an information system also called M2inder along with other datasets to probabilistically identify where and when structure (i.e., house) fires are likely to occur. This paper presents the findings from a participatory action research (PAR) study of the M2inder project, wherein a team of both practitioners and researchers with varying expertise, was assembled to guide the development of the M2inder system. Coupled with the M2inder application, a “deep learning” model was developed to enhance and validate the project. With the ability to predict the likelihood of a house fire event, the M2inder application could transform Emergency Management, permitting firefighting services to better prepare communities and strategically manage firefighting assets, locating them near the most likely predicted structure fires to minimize critical response times - a known key factor in saving lives and property. Interpreted through an activity theory lens, the findings of our research demonstrate how data mining capabilities of information systems can add ‘prediction’ to the traditional Comprehensive Emergency Management framework, thus revolutionizing firefighting service efficacy.