Abstract

Governments worldwide have adopted different strategies fronting the pandemic associated with the SARS-CoV-2. These measures include different approaches to assign and manage the resources, control the spread of the virus, and mitigate the contagious. The technical efficiency allows evaluating the success rate in government managing and health facilities performance. Since technical efficiency is a relative measure, experts must compare the hospitals by adjusting their production according to the type of patient treated: case-mix. The literature recommends using the Related Groups for Diagnosis system (DRG) to adjust hospitals' production. However, only 80 of the more than 195 public hospitals have implemented this system in Chile, limiting the evaluation of technical efficiency. The Ministry of Health of Chile (MINSAL) has proposed an administrative categorisation for the public hospitals: high, medium, and low complexity. Managers can use this classification to group the hospitals avoiding bias. However, how good is this classification according to the data science point of view? In this work, we evaluate the categorisation proposed by MINSAL by applying internal clustering indexes and input features associated with healthcare production related to case-mix, considering different clustering techniques and an ad-hoc NSGA-II strategy. After evaluating alternative partitions where the latter obtains the best quality scores, we propose a new classification for technical efficiency analyses.