Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

4TaStiC: Time and trend traveling time series clustering for classifying long-term type 2 diabetes patients

Onthada Preedasawakul, Nathakhun Wiroonsri

TL;DR

This paper tackles clustering of long-term HbA1c time series in type 2 diabetes patients despite irregular visit timings and evolving patterns. It introduces 4TaStiC, a dissimilarity measure that jointly performs time traveling and trend traveling by optimizing over time shifts up to $L$ and tilt angles in $E$, blending $d_{weu}$ and Pearson correlation with a tunable weight $\alpha$. The method, implemented in the FourTaStiC R package, is evaluated against seven baselines on artificial data and applied to 1,989 Siriraj diabetes patients, yielding clinically meaningful seven groups with distinct HbA1c trajectories and complication profiles. The approach advances patient stratification for personalized care and has potential applicability beyond healthcare to any domain with pattern-driven time series data.

Abstract

Diabetes is one of the most prevalent diseases worldwide, characterized by persistently high blood sugar levels, capable of damaging various internal organs and systems. Diabetes patients require routine check-ups, resulting in a time series of laboratory records, such as hemoglobin A1c, which reflects each patient's health behavior over time and informs their doctor's recommendations. Clustering patients into groups based on their entire time series data assists doctors in making recommendations and choosing treatments without the need to review all records. However, time series clustering of this type of dataset introduces some challenges; patients visit their doctors at different time points, making it difficult to capture and match trends, peaks, and patterns. Additionally, two aspects must be considered: differences in the levels of laboratory results and differences in trends and patterns. To address these challenges, we introduce a new clustering algorithm called Time and Trend Traveling Time Series Clustering (4TaStiC), using a base dissimilarity measure combined with Euclidean and Pearson correlation metrics. We evaluated this algorithm on artificial datasets, comparing its performance with that of seven existing methods. The results show that 4TaStiC outperformed the other methods on the targeted datasets. Finally, we applied 4TaStiC to cluster a cohort of 1,989 type 2 diabetes patients at Siriraj Hospital. Each group of patients exhibits clear characteristics that will benefit doctors in making efficient clinical decisions. Furthermore, the proposed algorithm can be applied to contexts outside the medical field.

4TaStiC: Time and trend traveling time series clustering for classifying long-term type 2 diabetes patients

TL;DR

This paper tackles clustering of long-term HbA1c time series in type 2 diabetes patients despite irregular visit timings and evolving patterns. It introduces 4TaStiC, a dissimilarity measure that jointly performs time traveling and trend traveling by optimizing over time shifts up to and tilt angles in , blending and Pearson correlation with a tunable weight . The method, implemented in the FourTaStiC R package, is evaluated against seven baselines on artificial data and applied to 1,989 Siriraj diabetes patients, yielding clinically meaningful seven groups with distinct HbA1c trajectories and complication profiles. The approach advances patient stratification for personalized care and has potential applicability beyond healthcare to any domain with pattern-driven time series data.

Abstract

Diabetes is one of the most prevalent diseases worldwide, characterized by persistently high blood sugar levels, capable of damaging various internal organs and systems. Diabetes patients require routine check-ups, resulting in a time series of laboratory records, such as hemoglobin A1c, which reflects each patient's health behavior over time and informs their doctor's recommendations. Clustering patients into groups based on their entire time series data assists doctors in making recommendations and choosing treatments without the need to review all records. However, time series clustering of this type of dataset introduces some challenges; patients visit their doctors at different time points, making it difficult to capture and match trends, peaks, and patterns. Additionally, two aspects must be considered: differences in the levels of laboratory results and differences in trends and patterns. To address these challenges, we introduce a new clustering algorithm called Time and Trend Traveling Time Series Clustering (4TaStiC), using a base dissimilarity measure combined with Euclidean and Pearson correlation metrics. We evaluated this algorithm on artificial datasets, comparing its performance with that of seven existing methods. The results show that 4TaStiC outperformed the other methods on the targeted datasets. Finally, we applied 4TaStiC to cluster a cohort of 1,989 type 2 diabetes patients at Siriraj Hospital. Each group of patients exhibits clear characteristics that will benefit doctors in making efficient clinical decisions. Furthermore, the proposed algorithm can be applied to contexts outside the medical field.
Paper Structure (24 sections, 3 theorems, 13 equations, 8 figures, 7 tables, 1 algorithm)

This paper contains 24 sections, 3 theorems, 13 equations, 8 figures, 7 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Overview and background
  3. Our proposed model overview
  4. Clustering methods
  5. K-means Clustering Kmeans1kmeans2
  6. Hierarchical clustering hc1hc2
  7. Cluster evaluation
  8. Existing distances and dissimilarity measures
  9. Our proposed methods
  10. Time traveling
  11. Trend traveling
  12. 4TaStiC dissimilarity measure
  13. 4TaStiC clustering algorithm
  14. Mathematical properties
  15. Experimental results
  16. ...and 9 more sections

Key Result

Proposition 3.1

Let $L<T$ be non-negative integers, and $0 \in E \subset \mathbb{R}$. Then where $d_{L,E}(\cdot,\cdot)$ and $d(\cdot,\cdot)$ are defined as in 4TaStiC and basedist, respectively.

Figures (8)

  • Figure 1: Examples of two pairs of patients' HbA1c time series before (left) and after time and trend traveling (right)
  • Figure 2: The 4TaStiC idea map for diabetes patients
  • Figure 3: A dataset of example 1
  • Figure 4: Artificial datasets
  • Figure 5: The first 10 rows of the dataset
  • ...and 3 more figures

Theorems & Definitions (4)

  • Proposition 3.1
  • Proposition 3.2
  • Proposition 3.3
  • Example 3.4