A hybrid intelligent classifier to estimate obesity levels based on ERG signals
| Autor: | Irem Senyer Yapici, Okan Erkaymaz, Rukiye Uzun Arslan |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Discrete wavelet transform
Physics genetic structures Artificial neural network business.industry General Physics and Astronomy Particle swarm optimization Pattern recognition Context (language use) 01 natural sciences eye diseases 010305 fluids & plasmas Wavelet Classification of obesity 0103 physical sciences Classifier (linguistics) sense organs Artificial intelligence 010306 general physics business Erg |
| Zdroj: | Physics Letters A. 399:127281 |
| ISSN: | 0375-9601 |
| DOI: | 10.1016/j.physleta.2021.127281 |
| Popis: | Obesity is a worldwide prevalence metabolic disease causing significant eye problems. Body Mass Index is proved to not be a sufficient criterion to classify obesity. In this context, a diagnostic support system for determining obesity levels by using electroretinogram signals is designed. To do this, the discrete wavelet transform is applied to three different electroretinogram responses recorded from both eyes. The obtained wavelet coefficients' size is reduced using statistical property. The designed dataset is used in artificial neural networks and artificial neural networks based particle swarm optimization models to classify obesity. We found that the average accuracy of the hybrid model is higher than the traditional model and the cone response is a highly effective response in obesity classification. This study is the first attempt to classify obesity levels based on electroretinogram signals and this study shows that obesity can be classified from electroretinogram signals. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect