Skip to main content

Table 1 Identification and performance of linear classification models. Classification of PI-CFS and healthy recovered control (RC) subjects at 6, 12 and 24 months post-diagnosis with IM using linear discriminant models with terms identified as optimal at each time point based on classical stepwise variable selection with the exception of classification based on glucose levels at 6 months which was specified based on differential expression

From: Tracking post-infectious fatigue in clinic using routine Lab tests

Model Assigned RC Assigned PI-FS Total   Accuracy NPV PPV Specificity Sensitivity
6 months          
 ACTH 8 4 12 True RC 0.72 0.73 0.71 0.67 0.77
  3 10 13 True PI-CFS      
 Glucose 9 3 12 True RC 0.72 0.69 0.75 0.75 0.69
  4 9 13 True PI-CFS      
12 months          
 Estradiol 9 3 12 True RC 0.76 0.75 0.77 0.75 0.77
  3 10 13 True PI-CFS      
24 months          
 Neutrophils 10 2 12 True RC 0.84 0.83 0.85 0.83 0.85
  2 11 13 True PI-CFS