Skip to main content
Fig. 4 | BMC Medical Informatics and Decision Making

Fig. 4

From: Identifying the presence and severity of dementia by applying interpretable machine learning techniques on structured clinical records

Fig. 4

The decision tree predicts the dementia classes (“No Dementia”/ “Minimal or Mild Dementia” / “Moderate or Severe Dementia”) of 2505 patients for the Problem (b). The pie charts in leaves show class labels, the proportion of resulting classes and their support size. The branches demonstrate connections between features and their threshold values, leading towards class labels. In the tree, some most significant paths A, B and C predict nearly pure decisive classes for patients. The rules of path A and lead to a pure “Moderate or Severe Dementia” population of 71 patients. The rules of paths B and C lead to a large majority of “Mild or Minimal Dementia” and “No Dementia” populations, comprised by 113 and 1111 individuals, respectively (see Additional file 2: Figure S2 in Supplementary Information for the better visibility)

Back to article page