Topological Analysis of Graphs for Comparing Sets of Time Series: A Case Study with Hospital Data

Analyzing multiple time series and their correlations is exceptionally challenging due to the complexity of the data, temporal variability, and the need for advanced techniques to identify patterns and predict future behaviors. This work presents a methodology for this purpose by analyzing data on the use of hospital resources at the Municipal Hospital of São José dos Campos before and during the COVID-19 pandemic. We extracted time-series meta-features to describe the time series of each resource and created graphs for each year based on the Euclidean distance between the features. By examining the structure of these graphs, we identified significant changes in the pattern of resource utilization during the pandemic. Our results show that the dissimilarity in consumption profiles between items increased, and the connectivity of the graphs decreased, highlighting the impact of the pandemic on hospital operations.