We analyze a dataset which measures the acceptance of EEG home- monitoring dependent on different variables, such as personnel characteristics, for example. A standard approach to model these relations is the logistic regression. Another one is an ANN (Artificial Neural Network). We used 70% of the dataset to determine parameters and the remaining 30% to compare predictions with outcomes and measure the performance of both approaches. Both methods result in good predictive success. Surprisingly, the logistic regression is slightly better than the ANN. The ordering of the variables according to their importance is different in both approaches (rank correlation coefficient of − 0.28). Therefore, the interpretation of the data crucially depends on the method. The use of these methods depends on which information is available: a full model as in the logistic regression or only the variables or the use of different optimization criteria in the approaches. We discuss this result.