Fig. 10From: Identifying the presence and severity of dementia by applying interpretable machine learning techniques on structured clinical recordsLIME provides the local interpretability for the prediction of an arbitrary test instance with prediction probabilities for Problem (b). The left part of the figure shows the weights of the top 10 important features while making class decision, whereas the right part represents predicted class probability and the top 10 important feature with their values. The weight of the feature represents its importance. Here, the short names of the impacting features in the left part of the figure are in the same order as in the feature & value list in the right part. Min-Mild Dem. and Mod-Sev Dem are the abbreviations for the Minimal or Mild dementia and Moderate or Severe dementia classes, respectivelyBack to article page