End-to-End Depression Recognition Based on a One-Dimensional Convolution Neural Network Model Using Two-Lead ECG Signal.

Purpose: Depression is a common mental illness worldwide and has become an important public health problem. The current clinical diagnosis of depression mainly relies on the doctor's experience and subjective diagnosis, which results in the low diagnostic efficiency and insufficient objectivity of diagnostic results. Therefore, establishing a physiological and psychological model for computer-aided diagnosis is an urgent task. In order to solve the above problems, this article uses a convolutional neural network (CNN) to identify depression based on electrocardiogram (ECG). Methods: Our method uses the raw ECG signal as the input of one-dimensional CNN, and uses the automatic feature processing layer of CNN to learn and distinguish signal features without additional feature extraction and feature selection steps. In order to obtain the optimal model, ECG segments of different durations (3 s, 4 s, 5 s and 6 s) and CNNs with different layers were used for comparison. In order to obtain modeling data, the resting ECG of 37 depression patients and 37 healthy controls were collected. In the proposed network, larger convolution kernels are used to better focus on overall changes. In addition, this article focuses on the inter-patient data classification standard, where the training and test sets come from different patient data. Results: Through comprehensive comparison, the 5 s ECG segment and 5-layer CNN are recommended in related applications. The proposed approach achieves high classification performance with accuracy of 93.96%, sensitivity of 89.43%, specificity of 98.49%, positive productivity of 98.34%. Conclusion: The experimental results indicate that the end-to-end deep learning approach can identify depression from ECG signals, and possess high diagnostic performance. It also shows that ECG is a potential biomarker in the diagnosis of depression.

Automated detection of clinical depression based on convolution neural network model.

As a common mental disorder, depression is placing an increasing burden on families and society. However, the current methods of depression detection have some limitations, and it is essential to find an objective and efficient method. With the development of automation and artificial intelligence, computer-aided diagnosis has attracted more and more attention. Therefore, exploring the use of deep learning (DL) to detect depression has valuable potential. In this paper, convolutional neural network (CNN) is applied to build a diagnostic model for depression based on electroencephalogram (EEG). EEG recordings are analyzed by three different CNN structures, namely EEGNet, DeepConvNet and ShallowConvNet, to dichotomize depression patients and healthy controls. EEG data were collected in the resting state from three electrodes (Fp1, Fz, Fp2) among 80 subjects (40 depressive patients and 40 normal subjects). After the preprocessing step, the DL structures are employed to classify the data, and their recognition performance is evaluated by comparing the classification results. The classification performance shows that depression was effectively detected using EEGNet with 93.74% accuracy, 94.85% sensitivity and 92.61% specificity. In the process of optimizing the parameters of EEGNet structure, the highest accuracy can reach 94.27%. Compared with traditional diagnostic methods, EEGNet is highly worthy for the future depression detection and valuable in terms of accuracy and speed.

Discrimination of vestibular function based on inertial sensors.

BACKGROUND AND OBJECTIVE: Vestibular dysfunction, as a common disease or symptom, can cause abnormalities in gait and balance. Since the existing detection methods are static detection and cannot obtain the dynamic vestibular information of patients, this paper proposes a simple method for detecting vestibular dysfunction based on gait signals of subjects. METHODS: In our study, the walking patterns of dynamic gait index (DGI) and inertial sensor were adopted for the data acquisition. Time-domain, frequency-domain and non-linear features were extracted from inertial sensor signals. Then the Relief algorithm was used for feature selection. Two classifiers, Support Vector Machine (SVM) and Random Forest (RF), were used to classify the patients with vestibular dysfunction and the healthy controls. RESULTS: The highest accuracy of 84.79% was achieved based on magnetometer features and SVM classifier. To further improve classification results, features of three sensor signals were combined and applied to two classifiers. Combined features and RF classifier achieved a classification accuracy of 86.5%. CONCLUSION: The detection of vestibular dysfunction based on inertial sensors might be simple, accurate and easy to implement in clinical examination, which provides a new method for the clinical diagnosis of vestibular function.