Deshmukh Manjusha, Khemchandani Mahi, Thakur Paramjit Mahesh
Computer Engineering Department, Saraswati College of Engineering, Mumbai, India.
Information Technology, Saraswati College of Engineering, Mumbai, India.
Appl Neuropsychol Child. 2024 Oct 1:1-15. doi: 10.1080/21622965.2024.2405719.
Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by repeated patterns of hyperactivity, impulsivity, and inattention that limit daily functioning and development. Electroencephalography (EEG) anomalies correspond to changes in brain connection and activity. The authors propose utilizing empirical mode decomposition (EMD) and discrete wavelet transform (DWT) for feature extraction and machine learning (ML) algorithms to categorize ADHD and control subjects. For this study, the authors considered freely accessible ADHD data obtained from the IEEE data site. Studies have demonstrated a range of EEG anomalies in ADHD patients, such as variations in power spectra, coherence patterns, and event-related potentials (ERPs). Some of the studies claimed that the brain's prefrontal cortex and frontal regions collaborate in intricate networks, and disorders in either of them exacerbate the symptoms of ADHD. , Based on the research that claimed the brain's prefrontal cortex and frontal regions collaborate in intricate networks, and disorders in either of them exacerbate the symptoms of ADHD, the proposed study examines the optimal position of EEG electrode for identifying ADHD and in addition to monitoring accuracy on frontal/ prefrontal and other regions of brain our study also investigates the position groupings that have the highest effect on accurateness in identification of ADHD. The results demonstrate that the dataset classified with AdaBoost provided values for accuracy, precision, specificity, sensitivity, and F1-score as 1.00, 0.70, 0.70, 0.75, and 0.71, respectively, whereas using random forest (RF) it is 0.98, 0.64, 0.60, 0.81, and 0.71, respectively, in detecting ADHD. After detailed analysis, it is observed that the most accurate results included all electrodes. The authors believe the processes can detect various neurodevelopmental problems in children utilizing EEG signals.
注意缺陷多动障碍（ADHD）是一种神经发育障碍，其特征是多动、冲动和注意力不集中的反复出现模式，这些模式会限制日常功能和发育。脑电图（EEG）异常与大脑连接和活动的变化相对应。作者建议利用经验模态分解（EMD）和离散小波变换（DWT）进行特征提取，并使用机器学习（ML）算法对ADHD患者和对照受试者进行分类。在本研究中，作者考虑了从IEEE数据网站获取的可免费访问的ADHD数据。研究表明，ADHD患者存在一系列EEG异常，如功率谱变化、相干模式和事件相关电位（ERP）。一些研究声称，大脑的前额叶皮质和额叶区域在复杂的网络中协同工作，其中任何一个区域出现紊乱都会加重ADHD的症状。基于声称大脑前额叶皮质和额叶区域在复杂网络中协同工作，且其中任何一个区域出现紊乱都会加重ADHD症状的研究，本研究考察了用于识别ADHD的EEG电极的最佳位置，除了监测额叶/前额叶和大脑其他区域的准确性外，我们的研究还调查了对ADHD识别准确性影响最大的位置分组。结果表明，使用AdaBoost分类的数据集在检测ADHD时，准确率、精确率、特异性、灵敏度和F1分数分别为1.00、0.70、0.70、0.75和0.71，而使用随机森林（RF）时分别为0.98、0.64、0.60、0.81和0.71。经过详细分析，发现最准确的结果包括所有电极。作者认为，这些过程可以利用EEG信号检测儿童的各种神经发育问题。
