A secondary analysis shows machine-learning models using routine clinical and cognitive variables achieved moderate-to-good ...