Skip to main content
. 2021 Jan 22;13(3):e13257. doi: 10.15252/emmm.202013257

Figure 6. Machine learning‐based classification of PD and LRRK2 status.

Figure 6

  1. Receiver operating characteristic (ROC) curve for the XGBoost‐based model to classify PD+ vs. PD− individuals. Random performance is indicated by the dotted diagonal line. The gray area represents the standard deviation from the mean ROC curve. The blue lines show the values for a total of 15 repeats with four stratified train‐test splits.
  2. Confusion matrix showing the model performance for classifying PD+ vs. PD− individuals. Numbers represent the mean number from 15 repeats of cross‐validation with four stratified train‐test splits.
  3. ROC curve for the XGBoost‐based model when trained on one cohort and tested on the other cohort. Random performance is indicated by the dotted diagonal line.
  4. Same as (A) but for classification of LRRK2 G2019S vs. LRRK2 WT carriers.
  5. Same as (B) but for classification of LRRK2 G2019S vs. LRRK2 WT carriers.
  6. Same as (C) but for classification of LRRK2 G2019S vs. LRRK2 WT carriers.
  7. Same as (A) but for classification of PD+ vs. PD− in LRRK2 G2019S carriers.
  8. Same as (B) but for classification of PD+ vs. PD− in LRRK2 G2019S carriers.
  9. Same as (C) but for classification of PD+ vs. PD− in LRRK2 G2019S carriers.