Fig. 2: The relative predictive importance ranking of clinical factors in random forest models fitted with binary variables indicating the presence or non-presence of elevated levels of each of the three CNS damage markers as response variables. | Translational Psychiatry

Fig. 2: The relative predictive importance ranking of clinical factors in random forest models fitted with binary variables indicating the presence or non-presence of elevated levels of each of the three CNS damage markers as response variables.

From: An exploratory study of the damage markers NfL, GFAP, and t-Tau, in cerebrospinal fluid and other findings from a patient cohort enriched for suspected autoimmune psychiatric disease

Fig. 2

A rank of 100 is the most important, and lower numbers showing the relative importance of different variables compared to the most important predictor. Where variables used in all models vs not used in the models at all are ranked 100 respective 0.

Back to article page