Discovering Symptom Co-Occurrence Patterns from 604 Cases of Depressive Patient Data Using Latent Tree Models

Objectives: In order to treat depressive patients using Traditional Chinese Medicine (TCM), it is necessary to classify them into subtypes from the TCM perspective. Those subtypes are called Zheng types. This article aims at providing evidence for the classification task by discovering symptom co-occurrence patterns from clinic data. Methods: Six hundred four (604) cases of depressive patient data were collected. The subjects were selected using the Chinese classification of mental disorder clinic guideline CCMD-3. The symptoms were selected based on the TCM literature on depression. The data were analyzed using latent tree models (LTMs). Results: An LTM with 29 latent variables was obtained. Each latent variable represents a partition of the subjects into 2 or more clusters. Some of the clusters capture probabilistic symptom co-occurrence patterns, while others capture symptom mutual-exclusion patterns. Most of the co-occurrence patterns have clear TCM Zheng connotations. Conclusions: From clinic data about depression, probabilistic symptom co-occurrence patterns have been discovered that can be used as evidence for the task of classifying depressive patients into Zheng types