Clustering of timed sequences -- Application to the analysis of care pathways

TL;DR

This paper introduces novel clustering algorithms for timed sequences, specifically care pathways, by adapting time series methods like drop-DTW and DBA, and evaluates their effectiveness on synthetic and real healthcare data.

Contribution

It adapts time series clustering techniques to timed sequences, addressing the challenge of defining meaningful metrics and algorithms for care pathway analysis.

Findings

Effective clustering of care pathways demonstrated on real-world data

Proposed methods outperform traditional clustering approaches

Validated on synthetic and actual healthcare datasets

Abstract

Improving the future of healthcare starts by better understanding the current actual practices in hospital settings. This motivates the objective of discovering typical care pathways from patient data. Revealing typical care pathways can be achieved through clustering. The difficulty in clustering care pathways, represented by sequences of timestamped events, lies in defining a semantically appropriate metric and clustering algorithms. In this article, we adapt two methods developed for time series to the clustering of timed sequences: the drop-DTW metric and the DBA approach for the construction of averaged time sequences. These methods are then applied in clustering algorithms to propose original and sound clustering algorithms for timed sequences. This approach is experimented with and evaluated on synthetic and real-world data.