Extracting most impacting emergency department patient flow during winter outbreak by analyzing and treating time series
Emergency departments (ED) in France are jeopardized each winter by the respiratory viruses and viral outbreak. This overcrowding badly impacts the hospital and induces a degradation of the welcome capacity and of the care quality. To avoid this phenomenon, action plans can be set up (increasing the number of beds, staff,..), but they have human and material costs. It is important to detect early and with precision when the overcrowding will occur. Those tools can be efficient only if a work on the comprehension of the patient flow in the ED is done beforehand. Indeed, the patient flow is due to a large variety of diagnoses and only a few are linked with winter epidemics.
To tackle this, we propose in this work to compare the laboratory-confirmed positive test of viruses and the time series of the diagnostic code patient. The laboratory-confirmed data takes time to be available, so we want the diagnostic code linked to those viruses in order to have a real time visualisazion of the impact of the viruses. With clustering methods, we match together similar behaviour.