IEEE Trans Nanobioscience. 2020 Jul;19(3):527-537. doi: 10.1109/TNB.2020.2990690. Epub 2020 Apr 27.
At present, depression has become a main health burden in the world. However, there are many problems with the diagnosis of depression, such as low patient cooperation, subjective bias and low accuracy. Therefore, reliable and objective evaluation method is needed to achieve effective depression detection. Electroencephalogram (EEG) and eye movements (EMs) data have been widely used for depression detection due to their advantages of easy recording and non-invasion. This research proposes a content based ensemble method (CBEM) to promote the depression detection accuracy, both static and dynamic CBEM were discussed. In the proposed model, EEG or EMs dataset was divided into subsets by the context of the experiments, and then a majority vote strategy was used to determine the subjects' label. The validation of the method is testified on two datasets which included free viewing eye tracking and resting-state EEG, and these two datasets have 36,34 subjects respectively. For these two datasets, CBEM achieves accuracies of 82.5% and 92.65% respectively. The results show that CBEM outperforms traditional classification methods. Our findings provide an effective solution for promoting the accuracy of depression identification, and provide an effective method for identificationof depression, which in the future could be used for the auxiliary diagnosis of depression.
目前，抑郁症已成为全球主要的健康负担。然而，抑郁症的诊断存在许多问题，如患者配合度低、主观偏差和准确率低等。因此，需要可靠和客观的评估方法来实现有效的抑郁症检测。脑电图（EEG）和眼动（EMs）数据由于易于记录和非侵入性的优点，已被广泛用于抑郁症检测。本研究提出了一种基于内容的集成方法（CBEM），以提高抑郁症检测的准确性，讨论了静态和动态 CBEM。在提出的模型中，EEG 或 EMs 数据集根据实验的上下文划分为子集，然后使用多数投票策略来确定受试者的标签。该方法的验证是在两个数据集上进行的，包括自由观看眼动追踪和静息态 EEG，这两个数据集分别有 36、34 个受试者。对于这两个数据集，CBEM 的准确率分别达到 82.5%和 92.65%。结果表明，CBEM 优于传统的分类方法。我们的研究结果为提高抑郁症识别的准确性提供了有效的解决方案，并为抑郁症的识别提供了一种有效的方法，未来可用于抑郁症的辅助诊断。
