Fig. 1

The workflow of the cohort selection, pre-processing, imputation, and ML model training steps. a Selected version and tables of Enroll-HD. b Inclusion criteria of our study. c Pre-processing steps for the reduction of the number of missing values in the cohort. d Imputation of the remaining missing values using ML models. e Prediction steps. For the AAO prediction two cohorts are created one for the narrow CAG size (41–56) and another for the wide CAG size (36–59) to fit and evaluate the ML models and the Langbehn formula. For the driving capability the GRUs are fitted and evaluated with multiple hyperparameters