Department of Child Healthcare, Cangzhou Central Hospital, Cangzhou 061000, Hebei, China.
J Healthc Eng. 2021 Nov 25;2021:8201363. doi: 10.1155/2021/8201363. eCollection 2021.
This study aimed to analyze the application of composite domain analysis algorithm for electroencephalogram (EEG) images of children with epilepsy and to investigate the risk factors related to poor prognosis. 70 children with neurological epilepsy admitted to the hospital were selected as the research objects. Besides, the EEG of the children during the intermittent and seizure phases of epilepsy were collected, so as to establish a composite domain analysis algorithm model. Then, the model was applied in EEG analysis. The clinical disease type and prognosis of children were statistically analyzed, and the risk factors that affected the prognosis of children were investigated. The results showed that the EEG signal values of the detail coefficients (d51 and d52) and the approximate coefficient (c5) during the epileptic seizure period were higher markedly than the signal values of the epileptic intermittent period; the EEG signal of the epileptic intermittent period was a transient waveform, which appeared as sharp waves or spikes. The EEG signal of epileptic seizures was continuous, with a composite waveform of sharp waves and spikes, and the change amplitude of the wavelet envelope spectrum during epileptic seizures was also higher hugely than that of intermittent epilepsy. The accurate identification rate, specificity, and sensitivity of EEG analysis with the composite domain algorithm were higher than those without the algorithm. Among the five types of epileptic seizures in children, the proportion of systemic tonic-clonic status was the largest, and the proportion of myoclonic status was equal to that of complex partial epileptic status, both of which were relatively small. The proportion of children with a better prognosis was 75.71% (53/70), which was higher than those with a poor prognosis 24.29% (17/70). Abnormal imaging examination (odds ratio (OR) = 3.823 and 95% confidence interval (CI) = 1.643-8.897); seizure duration greater than 1 hour (OR = 1.855 and 95% CI = 1.076-3.199); C-reactive protein (CRP) (OR = 5.089 and 95% CI = 1.507-17.187); and abnormal blood glucose (OR = 3.077, 95%CI = 1.640-5.773) were all independent risk factors for poor prognosis (all < 0.05). The composite domain analysis algorithm was helpful for clinicians to find the difference in the EEG signals between the epileptic seizure period and the epileptic intermittent period in a short time, thereby improving the doctor's analysis of the results, which could reflect its marked superiority. In addition, abnormal imaging examinations, convulsion duration greater than 1 hour, CRP, and abnormal blood glucose were independent risk factors for poor prognosis in children. Therefore, the invasion of related risk factors could be reduced clinically by prognostic review with medical advice, attention to food safety and hygiene, and improvement of children's immunity.
本研究旨在分析复合域分析算法在儿童癫痫脑电图（EEG）图像中的应用，并探讨与预后不良相关的危险因素。选择 70 名因神经系统癫痫住院的儿童作为研究对象。此外，收集儿童癫痫间歇性和发作期间的 EEG，建立复合域分析算法模型。然后，将模型应用于 EEG 分析。统计分析儿童的临床疾病类型和预后，并探讨影响儿童预后的危险因素。结果表明，癫痫发作期间细节系数（d51 和 d52）和近似系数（c5）的 EEG 信号值明显高于癫痫间歇性期间的信号值；癫痫间歇性期间的 EEG 信号是一种瞬态波形，表现为尖波或棘波。癫痫发作期间的 EEG 信号是连续的，具有尖波和棘波的复合波形，并且癫痫发作期间的小波包谱的变化幅度也大大高于癫痫间歇性。具有复合域算法的 EEG 分析的准确识别率、特异性和敏感性均高于没有算法的分析。在儿童的五种癫痫发作类型中，全身性强直-阵挛状态的比例最大，肌阵挛状态的比例与复杂部分癫痫状态相等，均相对较小。预后较好的儿童比例为 75.71%（53/70），高于预后较差的儿童 24.29%（17/70）。异常影像检查（比值比（OR）=3.823，95%置信区间（CI）=1.643-8.897）；发作持续时间大于 1 小时（OR=1.855，95%CI=1.076-3.199）；C 反应蛋白（CRP）（OR=5.089，95%CI=1.507-17.187）；以及异常血糖（OR=3.077，95%CI=1.640-5.773）均为预后不良的独立危险因素（均<0.05）。复合域分析算法有助于临床医生在短时间内找到癫痫发作期和癫痫间歇性期 EEG 信号的差异，从而提高医生对结果的分析能力，这凸显了其明显的优势。此外，异常影像检查、发作持续时间大于 1 小时、CRP 和异常血糖是儿童预后不良的独立危险因素。因此，临床上可通过预后评估、医疗建议、注意食品安全和卫生以及提高儿童免疫力来减少相关危险因素的侵袭。
