Artificial Intelligence Based Performance Analysis in Detection of Schizophrenia

Ali Narin; Burhan Sezen

Makale detayları

Artificial Intelligence Based Performance Analysis in Detection of Schizophrenia

Yazarlar:

Ali Narin | Burhan Sezen

Yayıncı:

Karyay Karadeniz Yayımcılık

Ana Başlıklar:

Biyomedikal Tanı

Dergi:

Black Sea Journal of Engineering and Science Vol.8 Issue 3, pp.695-703

Belge Tipi:

Makale

Yayın Tarihi:

5/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001345284

pISSN:

2619-8991

Anahtar kelimeler:

Özet:

 Schizophrenia SZ is a mental illness in which there are negative perceptions and behaviors, and it is not possible to distinguish between real and imaginary sounds or experiences. Hallucinations, delusions, difficulty in speaking and expressing emotions are among the symptoms of schizophrenia. Early detection of SZ is of vital importance in order to prevent such permanent problems. Blood tests, brain imaging and EEG signals are mostly used together with preliminary examination for diagnosis. Since EEG has a very low amplitude and contains very different frequencies, SZ detection performance may vary. In this study, SZ patients will be detected using EEG signals. The study results were obtained from EEG data of 39 healthy and 45 SZ patients. The performance of the detected features was recorded by applying backward elimination method together with time and frequency domain measurements on different normalized EEG signals. The k-nearest neighbor k-NN algorithm was used in the study. As a result, 83.6 accuracy value was obtained for k-NN k1 with z-score normalization and 13 features. These results show that SZ patients can be detected with very high accuracy. It is also thought to be useful as a decision support system for neurology specialists.

 Schizophrenia SZ is a mental illness in which there are negative perceptions and behaviors, and it is not possible to distinguish between real and imaginary sounds or experiences. Hallucinations, delusions, difficulty in speaking and expressing emotions are among the symptoms of schizophrenia. Early detection of SZ is of vital importance in order to prevent such permanent problems. Blood tests, brain imaging and EEG signals are mostly used together with preliminary examination for diagnosis. Since EEG has a very low amplitude and contains very different frequencies, SZ detection performance may vary. In this study, SZ patients will be detected using EEG signals. The study results were obtained from EEG data of 39 healthy and 45 SZ patients. The performance of the detected features was recorded by applying backward elimination method together with time and frequency domain measurements on different normalized EEG signals. The k-nearest neighbor k-NN algorithm was used in the study. As a result, 83.6 accuracy value was obtained for k-NN k1 with z-score normalization and 13 features. These results show that SZ patients can be detected with very high accuracy. It is also thought to be useful as a decision support system for neurology specialists.

 Schizophrenia SZ is a mental illness in which there are negative perceptions and behaviors, and it is not possible to distinguish between real and imaginary sounds or experiences. Hallucinations, delusions, difficulty in speaking and expressing emotions are among the symptoms of schizophrenia. Early detection of SZ is of vital importance in order to prevent such permanent problems. Blood tests, brain imaging and EEG signals are mostly used together with preliminary examination for diagnosis. Since EEG has a very low amplitude and contains very different frequencies, SZ detection performance may vary. In this study, SZ patients will be detected using EEG signals. The study results were obtained from EEG data of 39 healthy and 45 SZ patients. The performance of the detected features was recorded by applying backward elimination method together with time and frequency domain measurements on different normalized EEG signals. The k-nearest neighbor k-NN algorithm was used in the study. As a result, 83.6 accuracy value was obtained for k-NN k1 with z-score normalization and 13 features. These results show that SZ patients can be detected with very high accuracy. It is also thought to be useful as a decision support system for neurology specialists.

 Şizofreni SZ, algıda ve davranışlarda olumsuzlukların yaşandığı, seslerin ya da yaşananların gerçek veya hayal ayrımının yapılamadığı zihinsel bir hastalıktır. Halüsinasyonlar, hezeyanlar, konuşma ve duyguyu ifade etmede zorluk yaşama şizofreni belirtileri arasında yer almaktadır. Bu gibi kalıcı problemlerin önüne geçilmesi için SZnin erken tespiti hayati öneme sahiptir. Tanı amaçlı çoğunlukla ön muayene ile birlikte kan tahlili, beyin görüntüleme ve elektroensefalogram EEG sinyalleri kullanılmaktadır. EEG oldukça düşük genliğe sahip olmasının yanı sıra çok farklı frekanslar barındırması nedeniyle SZnin tespit performansı değişebilmektedir. Bu çalışmada EEG sinyalleri kullanılarak SZ hastalarının tespiti gerçekleştirilecektir. 39 sağlıklı, 45 SZ hastasına ait EEG verileri üzerinden çalışma sonuçları elde edilmiştir. Farklı normalize edilmiş EEG işaretleri üzerinden zaman ve frekans alanı ölçümleri ile birlikte geriye doğru eleme yöntemi de uygulanarak tespit edilen özniteliklerin başarımları kayıt altına alınmıştır. Çalışmada k-en yakın komşu k-NN algoritması kullanılmıştır. Sonuç olarak z-skor normalizasyonlu ve 13 öznitelikle k-NN k1 için 83,6 doğruluk değeri elde edilmiştir. Elde edilen bu sonuçlar SZ hastalarının oldukça yüksek başarım ile tespit edilebileceğini göstermektedir. Çalışmanın nöroloji uzmanlarına da karar destek sistemi olarak fayda vereceği düşünülmektedir.

Makale detayları

Artificial Intelligence Based Performance Analysis in Detection of Schizophrenia

Yazarlar:

Ali Narin | Burhan Sezen

Yayıncı:

Karyay Karadeniz Yayımcılık

Ana Başlıklar:

Biyomedikal Tanı

Dergi:

Black Sea Journal of Engineering and Science Vol.8 Issue 3, pp.695-703

Belge Tipi:

Makale

Yayın Tarihi:

5/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001345284

pISSN:

2619-8991

Schizophrenia

EEG

Feature extraction

Backward elimination

k-NNŞizofreni

EEG

Öznitelik çıkarımı

Geriye doğru eleme

k-NN

Sayı

Makale detayları

Yazarlar: Ali Narin | Burhan Sezen

Yayıncı: Karyay Karadeniz Yayımcılık

Ana Başlıklar: Biyomedikal Tanı

Dergi: Black Sea Journal of Engineering and Science Vol.8 Issue 3, pp.695-703

Belge Tipi: Makale

Yayın Tarihi: 5/2025

Belge Dili: Türkçe

Document ID (Org No.): 2001345284

pISSN: 2619-8991

Sayı

Yazarlar:

Ali Narin | Burhan Sezen

Yayıncı:

Karyay Karadeniz Yayımcılık

Ana Başlıklar:

Biyomedikal Tanı

Dergi:

Black Sea Journal of Engineering and Science Vol.8 Issue 3, pp.695-703

Belge Tipi:

Makale

Yayın Tarihi:

5/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001345284

pISSN:

2619-8991