Table 1.
Variables included in the ML models.
Table 2.
Distribution of binarization for each of the 10 target outcomes in the dataset.
Fig 1.
Optimal model performance metrics across the different outcome measures.
Fig 2.
Receiver operating characteristic of the best performing algorithm for the five most prevalent outcome measures in the dataset.
Fig 3.
Precision-recall curve of the best performing algorithm for the five most prevalent outcome measures in the dataset.
Table 3.
Optimal model performance and algorithm type for each of the ten outcome variables.
Fig 4.
Worked example of ML model for global HRQoL.
Twelve demographic and perioperative data inputs are used by the Support Vector Machine to predict whether global HRQoL will decline below the normative population mean (4) within 12–60 months of tumour resection.