Data Fusion in Forecasting Medical Demands based on Spectrum of Post-Earthquake Diseases

Abstract

Industry 4.0 makes it possible to develop smart emergency rescue systems in natural disasters. One of the most critical challenges is forecasting the demands of resources for appropriate resource allocations based on data from multiple sources with different levels of reliability. This paper deals with the challenge of data fusion and processing in forecasting resource demands for emergency responses to patients with various disease types. After an earthquake, the data on injuries, damages, and medical demands are characterized as diversified, unorganized, distributed, dynamic, and chaotic. Therefore, how to collect, filter, fuse, and mine data is most critical to forecast and allocate resources, especially for some emergent sources such as drugs for injuries and illnesses in post-earthquakes. To determine general patterns of outbreak diseases and corresponding medical needs, multi-source data is fused and processed to determine a reliable and accurate spectrum of post-earthquake diseases. The entropy-based weighting technology is adopted to determine the reliability and accuracy of data; the fused data is further processed to estimate the numbers of injuries, classify disease types, and finally predict the demands of medical supplies over time. In emergency rescues, medical resources are allocated and dispatched based on estimated numbers, types, and locations of patients. The effectiveness of the proposed method is verified and validated in simulation.