Fig. 3.

The validation of the binary classifiers of ALS using an unseen dataset. (A) The program setup for validating the binary classifier of ALS. (B) The performance of the four established models. This figure details the validation process of the binary classifiers designed to distinguish ALS samples from controls, using an independent dataset that was not included in the model-building phase. This “unseen dataset” is critical for evaluating how well the classifiers generalize to new data and ensuring that their performance holds up outside the training environment. The unseen dataset contains samples that the models have not encountered before, which allows us to assess the true predictive power of the classifiers in real-world scenarios. This validation process is a crucial step in confirming that the binary classifiers are not overfitted to the training data and can reliably generalize to new, unseen datasets.