OP-HHO based feature selection improves the performance of depression classification framework: A gender biased multiband research

被引:0
作者
Li, Kun [1 ]
Zhong, Peiyun [1 ]
Dong, Li [1 ]
Wang, Lingmin [1 ]
Jiang, Luo-Luo [2 ]
机构
[1] Guizhou Normal Univ, Sch Life Sci, Guiyang 550025, Guizhou, Peoples R China
[2] Zhejiang Univ Finance & Econ, Sch Informat Technol & Artificial Intelligence, Hangzhou 310018, Zhejiang, Peoples R China
基金
中国国家自然科学基金;
关键词
EEG; Depression; Optimized-parameter HHO; KNN; FRONTAL ALPHA ASYMMETRY; RESTING-STATE EEG; CLASSIFYING DEPRESSION; AUTOMATED DIAGNOSIS; BINARY PATTERN; SLEEP; DISORDER; EPILEPSY; THERAPY; POWER;
D O I
10.1016/j.amc.2025.129317
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Depression, as a common yet severe mood disorder, can cause irreversible damage to the brain if not detected and treated in a timely manner. Unfortunately, due to the current limitations of medical and technological conditions, only a small number of patients have been able to receive appropriate treatment. Although the traditional Harris Hawk's Optimization (HHO) algorithm has a strong searching ability for global optima which is helpful of early diagnosis of depression, it is highly prone to getting stuck in local optima during the early iterations. In view of this, the Optimized-Parameter Harris Hawk's Optimization (OP-HHO) algorithm proposed in this study is devised by integrating an exponential decay function. This incorporation endows the algorithm with the capacity to dynamically modulate the search step size, progressively diminish the escape energy, thereby bolstering the local search capabilities and efficaciously circumventing the problem of premature convergence that may stem from overzealous global exploration. The performance of the OP-HHO was tested using 23 benchmark functions. Based on the features selected by the OP-HHO algorithm, depression classification was carried out using the K-Nearest Neighbor (KNN) algorithm in combination with the MODMA database. The accuracy rate reached 96.36%- 97.30% across different brain wave frequencies under happy stimuli, and 100% under sad stimuli. Moreover, the classification results in the overall electroencephalogram (EEG) signals also showed excellent performance. This indicates that the OP-HHO algorithm is highly effective in accurately identifying the key features of depression. Our comparative study conducted reveals the existence of gender differences, which are expected to serve as effective features to further improve the accuracy of depression classification, opening up new avenues for the development of depression diagnosis techniques.
引用
收藏
页数:18
相关论文
共 76 条
[1]   Characterization of focal EEG signals: A review [J].
Acharya, U. Rajendra ;
Hagiwara, Yuki ;
Deshpande, Sunny Nitin ;
Suren, S. ;
Koh, Joel En Wei ;
Oh, Shu Lih ;
Arunkumar, N. ;
Ciaccio, Edward J. ;
Lim, Choo Min .
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2019, 91 :290-299
[2]   Automated EEG analysis of epilepsy: A review [J].
Acharya, U. Rajendra ;
Sree, S. Vinitha ;
Swapna, G. ;
Martis, Roshan Joy ;
Suri, Jasjit S. .
KNOWLEDGE-BASED SYSTEMS, 2013, 45 :147-165
[3]   Delta sleep EEG in depressed adolescent females and healthy controls [J].
Armitage, R ;
Emslie, GJ ;
Hoffmann, RF ;
Rintelmann, J ;
Rush, AJ .
JOURNAL OF AFFECTIVE DISORDERS, 2001, 63 (1-3) :139-148
[4]   EEG alpha asymmetry as a gender-specific predictor of outcome to acute treatment with different antidepressant medications in the randomized iSPOT-D study [J].
Arns, Martijn ;
Bruder, Gerard ;
Hegerl, Ulrich ;
Spooner, Chris ;
Palmer, Donna M. ;
Etkin, Amit ;
Fallahpour, Kamran ;
Gatt, Justine M. ;
Hirshberg, Laurence ;
Gordon, Evian .
CLINICAL NEUROPHYSIOLOGY, 2016, 127 (01) :509-519
[5]   Frontal and rostral anterior cingulate (rACC) theta EEG in depression: Implications for treatment outcome? [J].
Arns, Martijn ;
Etkin, Amit ;
Hegerl, Ulrich ;
Williams, Leanne M. ;
DeBattista, Charles ;
Palmer, Donna M. ;
Fitzgerald, Paul B. ;
Harris, Anthony ;
deBeuss, Roger ;
Gordon, Evian .
EUROPEAN NEUROPSYCHOPHARMACOLOGY, 2015, 25 (08) :1190-1200
[6]   Methods for classifying depression in single channel EEG using linear and nonlinear signal analysis [J].
Bachmann, Maie ;
Paeske, Laura ;
Kalev, Kaia ;
Aarma, Katrin ;
Lehtmets, Andres ;
Oopik, Pille ;
Lass, Jaanus ;
Hinrikus, Hiie .
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE, 2018, 155 :11-17
[7]  
[白露 BAI Lu], 2005, [中国心理卫生杂志, Chinese Mental Health Journal], V19, P719
[8]   The comparative effectiveness of electroencephalographic indices in predicting response to escitalopram therapy in depression: A pilot study [J].
Baskaran, Anusha ;
Farzan, Faranak ;
Milev, Roumen ;
Brenner, Colleen A. ;
Alturi, Sravya ;
McAndrews, Mary Pat ;
Blier, Pierre ;
Evans, Ken ;
Foster, Jane A. ;
Frey, Benicio N. ;
Giacobbe, Peter ;
Lam, Raymond W. ;
Leri, Francesco ;
MacQueen, Glenda M. ;
Muller, Daniel J. ;
Parikh, Sagar V. ;
Rotzinger, Susan ;
Soares, Claudio N. ;
Strother, Steven C. ;
Turecki, Gustavo ;
Kennedy, Sidney H. .
JOURNAL OF AFFECTIVE DISORDERS, 2018, 227 :542-549
[9]   Parkinson's disease: Cause factors, measurable indicators, and early diagnosis [J].
Bhat, Shreya ;
Acharya, U. Rajendra ;
Hagiwara, Yuki ;
Dadmehr, Nahid ;
Adeli, Hojjat .
COMPUTERS IN BIOLOGY AND MEDICINE, 2018, 102 :234-241
[10]   Differential neural response to psychoanalytic intervention techniques during structural interviewing: A single-case analysis using EEG [J].
Buchheim, Anna ;
Kernberg, Otto F. ;
Netzer, Nikolaus ;
Buchheim, Peter ;
Perchtold-Stefan, Corinna ;
Sperner-Unterweger, Barbara ;
Beckenbauer, Fabian ;
Labek, Karin .
FRONTIERS IN HUMAN NEUROSCIENCE, 2023, 16