Table 3 Confusion matrix for the clustering outcome.

From: An unsupervised XAI framework for dementia detection with context enrichment

Clusters/ disease diagnosis

CN

MCI

AD

FTD

Stable cluster

(N = 2084)

1383

(66.4%)

612

(29.4%)

63

(3%)

26

(1.2%)

Converter cluster

(N = 1318)

219

(16.6%)

503

(38.2%)

462

(35.1%)

134

(10.2%)

  1. The proportion of the participant’s baseline disease diagnosis stage and subtype within the found clusters - termed stable or converter - are presented as percentage points. CN cognitively normal, MCI mild cognitive impairment, AD dementia due to Alzheimer’s disease (the typical presentation), FTD Frontotemporal dementia where phenotypes include behavioral variant, semantic variant, and progressive nonfluent aphasia.
Back to article page