Identifying the Optimal Location to Set up an Emergency Room Using Machine Learning Algorithms

Abstract

: As per the the Allstate Canada Safe Driving Study report, Toronto was in 69 th position in accident rate among Canadian cities with an average of 6.45 accidents per 100 cars in 2014-2015. North York and Ajax, the borders of Toronto the accident rate was even worse with an average of 7.02 to 7.12 per 100 cars [1]. The higher the accident rate, the higher the death rate. By reducing the time lag between the accident and the initiation of medical care, one can prevent death or permanent disability. The distance between the accidents zones the emergency room play the vital role in reducing the death rate due to accident. As per the report, most of the accidents were at the outskirts of the city rather than within the city. But usually the most of the emergency rooms are within the city. In such cases mostly, the emergency rooms were far away from the accident zones. The objective of the work is to predict the most suitable place for establishing the emergency rooms using machine learning algorithms. Accident zones in Toronto, the dataset was taken from Toronto public service data portal and locations of emergency rooms were retrieved from Foursquare API. After mapping both the data set, the accident zones near the emergency rooms (which are at the distance of 1 km) are removed. Then accident dense area was found using hierarchical dbscan. K nearest neighbor algorithm is used to address the outliers. The suitable (core) location for the emergency room was found by taking the mean of each cluster. The distance between the core location and the emergency room was found. The core location with the longest distance was considered as the best place for establishing the new emergency room.