Extended Data Fig. 2: Workflow for machine learning analyses. | Nature Biomedical Engineering

Extended Data Fig. 2: Workflow for machine learning analyses.

From: Automated disc device for multiplexed extracellular vesicle isolation and labelling from liquid biopsies in cancer diagnostics

Extended Data Fig. 2: Workflow for machine learning analyses.

(a) Constructing a model for cancer diagnosis. The LASSO regression determined the most informative marker sets from the training set (n = 154; left). The selected markers were used to train an SVM model for binary classification (cancer versus non-cancer). This trained model was evaluated using an independent test set (n = 67; right). (b) Model for five-cancer classification. Data from all cancer patients (n = 157) were used. The model performed a one-versus-one classification to differentiate between five tumor types.

Back to article page