Comprehensive Summary
Electroencephalogram (EEG) signal analysis requires complex models to be able to analyze the unique signals that are present in an EEG when performing a clinical diagnosis. This study uses a Forest-based classification model which focuses on processing the EEG data, then extracting the features, followed by using an XGBoost-based feature selection algorithm and then deep forest classification. XGBoost is able to identify the most important features on an EEG using a comparison model and the deep forest model is able to address challenges in the EEG that can be due to noisy data. The scientists used an acquired dataset and a benchmark dataset to test the model and were able to generate feature columns using statistical and scientific equations. The accuracy score using this new model was 99.721% and the deep forest model is able to lower the computational time, which can be extremely beneficial. The scientists proved that a model using XGBoost and deep forest are effective when analyzing EEG signals raw data.
Outcomes and Implications
This research is important because EEG data is hard to understand and can have a considerable amount of noise, which makes it hard to determine which signals are most important. There have been other models used to analyze EEG signals, however, there are ways to make the models more accurate and efficient. Using computational models to help differentiate between signals would help to diagnose patients and save providers time. EEG signals are able to diagnose neurological conditions, which is why it is important for providers to be able to analyze the brain's electrical activity and see if there are any disturbances.