Optimizing Epileptic Seizure Recognition with Machine Learning Algorithms
According to the WHO, Epilepsy is a significant public health issue and increases every year from 1% to 2% in all age groups. It is one of the oldest recognized neurological disorders. Early detection and proper medication reduce the risk to the person. EEG is one of the methods to identify epilepsy...
Uloženo v:
| Vydáno v: | Journal of information systems engineering & management Ročník 10; číslo 5s; s. 543 - 557 |
|---|---|
| Hlavní autor: | |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
24.01.2025
|
| ISSN: | 2468-4376, 2468-4376 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | According to the WHO, Epilepsy is a significant public health issue and increases every year from 1% to 2% in all age groups. It is one of the oldest recognized neurological disorders. Early detection and proper medication reduce the risk to the person. EEG is one of the methods to identify epilepsy, The continuous monitoring of EEG signals recognizes seizures. These occur in the partial or total body or several parts of a brain for a person and it causes unconsciousness. A person who suffers from any one of the following health problems: high fever, sleepless nights, anxiety, and stress might cause epileptic seizures. Surviving with epilepsy is stressful and limited to employability. This work aims to build the best model using Machine Learning algorithms with high performance and accuracy values, by using the conventional Machine Learning algorithms like K- Nearest Neighbor, Support Vector Classifiers, Support Vector Regression, Lasso, Ridge, Decision Tree, Gradient Boost, eXtreme Gradient Boosting, Light Gradient Boosting Machine, Categorical Boosting, and Linear Regression. Optuna used for tuning hyperparameters in the ML models to improve the performance of a model and to obtain best results. The Kaggle data set used for training and validation purposes. In these models, according to classifiers models, all the “gradient boost” classifiers produce accuracy, precision, recall and F1-score with 1.0 values. In Regression models, the best model is “Linear Regression” with R2 score as 1.0 to detect epileptic seizures. |
|---|---|
| ISSN: | 2468-4376 2468-4376 |
| DOI: | 10.52783/jisem.v10i5s.682 |