Figure 4
From: Chronic back pain sub-grouped via psychosocial, brain and physical factors using machine learning
Confusion matrix of classifiers on chronic back pain and pain-free classes with (a) Support Vector Machine, (b) Naïve Bayes, (c) k-Nearest Neighbour and (d) Random Forest classifiers. The x-axis is the predicted class while the y-axis is the true class. Blue squares indicate the number in the class that was accurately classified, while the oranges squares show the number of misclassifications. The boxes on the right of the matrix show the percentage of classification (blue) and misclassification (orange) for the class. Classes are (0) pain-free individuals (n = 14,927; 78.2%), (1) low social isolation/loneliness and moderate depressive symptoms (n = 776; 4.1%), (2) low loneliness/social isolation and depressive symptoms (n = 2296; 12.0%), (3) high loneliness/social isolation and moderate depressive symptoms (n = 185; 1.0%), (4) moderate loneliness/social isolation and high depressive symptoms (n = 297; 1.6%) and (5) moderate loneliness/social isolation and low depressive symptoms (n = 602; 3.2%).
