Table 5 Cross-validated performance for integrative prediction of comorbidities, non-motor outcomes, and progression rate subgroups

From: Integrating digital gait data with metabolomics and clinical data to predict outcomes in Parkinson’s disease

Outcome Median AUC (mad)

Integrative Model Median AUC (mad)

Best individual data source Median AUC (mad)

Cognitive impairment (MoCA)

0.694 (0.12)

a0.788 (0.08)

Dopamine dysregulation syndrome

a0.857 (0.01)

0.714 (0.32)

Depression (BDI)

0.759 (0.06)

a0.783 (0.14)

Hallucinations

a0.813 (0.23)

0.785 (0.08)

Dyskinesias

0.901 (0.14)

a0.917 (0.12)

Apathy (Starkstein)

0.528 (0.26)

a0.616 (0.15)

Quality of life (PDQ-39)

a0.690 (0.12)

0.676 (0.21)

Progression rate

0.667 (0.16)

a0.728 (0.13)

  1. Cross-validated performance for predicting comorbidities, non-motor outcomes, and progression rate subgroups (left column) in PD by integrating time series features from raw signal gait data, clinical variables (non-motor features from the “Clinical data” section in the Methods), and blood metabolomics data. Quantitative outcome scores were binarized using a median threshold for comparable performance scores (for the progression rate outcome only, the fast and slow progression subgroups were defined as the top and bottom quartiles, respectively, of the average annual change in the MDS-UPDRS III motor score, consistent with previous studies27). The extreme gradient boosting (XGB) algorithm was used for prediction with 10-fold cross-validation. The presented scores represent the median area under the Receiver Operating Characteristic Curve (AUC). Column 1 shows the different outcome measures, column 2 the median AUC scores and median absolute deviations (mad) across the cross-validation cycles for the integrative models, and column 3 the AUC scores and standard deviations for the best models based on individual data sources (see Table 4).
  2. aThe highest median AUC.
Back to article page