Major depressive disorder prediction using data science

Background@#Major depressive disorder is a mood disorder that has affected many people worldwide. It is characterized by persistently low or depressed mood, anhedonia or decreased interest in pleasurable activities, feelings of guilt or worthlessness, lack of energy, poor concentration, appetite changes, psychomotor retardation or agitation, sleep disturbances, or suicidal thoughts. @*Objective@#The objective of the study was to predict the presence of major depressive disorder using a variety of machine learning classification algorithms (logistic regression, Naive Bayes, support vector machine, random forest, adaptive boosting, and extreme gradient boosting) on a publicly available depression dataset. @*Methodology@#After data pre-processing, several experiments were performed to assess the recursive feature elimination with cross validation as a feature selection method and synthetic minority over-sampling technique to address dataset imbalance. Several machine learning algorithms were applied on an anonymized publicly available depression dataset. Feature importance of the top performing models were also generated. All simulation experiments were implemented via Python 3.8 and its machine learning libraries (Scikit-learn, Keras, Tensorflow, Pandas, Matplotlib, Seaborn, NumPy). @*Results@#The top performing model was obtained by logistic regression with excellent performance metrics (91% accuracy, 93% sensitivity, 85% specificity, 93% recall, 93% F1-score, and 0.78 Matthews correlation coefficient). Feature importance scores of the most relevant attribute were also generated for the best model. @*Conclusion@#The findings suggest the utility of data science techniques powered by machine learning models to make a diagnosis of major depressive disorders with acceptable results. The potential deployment of these machine learning models in clinical practice can further enhance the diagnostic acumen of health professionals. Using data analytics and machine learning, data scientists can have a better understanding of mental health illness contributing to prompt and improved diagnosis thereby leading to the institution of early intervention and medical treatments ensuring the best quality of care for our patients.

Electrocardiogram signal classification algorithm of nested long short-term memory network based on focal loss function / 生物医学工程学杂志

Electrocardiogram (ECG) can visually reflect the physiological electrical activity of human heart, which is important in the field of arrhythmia detection and classification. To address the negative effect of label imbalance in ECG data on arrhythmia classification, this paper proposes a nested long short-term memory network (NLSTM) model for unbalanced ECG signal classification. The NLSTM is built to learn and memorize the temporal characteristics in complex signals, and the focal loss function is used to reduce the weights of easily identifiable samples. Then the residual attention mechanism is used to modify the assigned weights according to the importance of sample characteristic to solve the sample imbalance problem. Then the synthetic minority over-sampling technique is used to perform a simple manual oversampling process on the Massachusetts institute of technology and Beth Israel hospital arrhythmia (MIT-BIH-AR) database to further increase the classification accuracy of the model. Finally, the MIT-BIH arrhythmia database is applied to experimentally verify the above algorithms. The experimental results show that the proposed method can effectively solve the issues of imbalanced samples and unremarkable features in ECG signals, and the overall accuracy of the model reaches 98.34%. It also significantly improves the recognition and classification of minority samples and has provided a new feasible method for ECG-assisted diagnosis, which has practical application significance.

Development of a Portable Chronic Non-specific Low Back Pain Measurement System / 中国医疗器械杂志

We developed a portable non-specific low back pain measurement system EasiLBP and evaluated its performance in collecting EMG signals:during the wearer's movement without the assistance of a doctor, the collection of EMG signals by portable devices met problems such as large noise interference, difficulty in accurately calibrating the start and end points of the action interval, and imbalanced samples for feature recognition, et al. To challenge these problems, we proposed a small group-based noise removal method, a dynamic dual-threshold automatic method for identifying the start and end points of the motion interval, and a sampling method to balance group samples, respectively. Portable device and a medical EMG acquisition equipment Thought Technology FlexComp Infiniti 10 were used to perform EMG measurements on 15 patients with non-specific low back pain and 15 normal people. Clinical experiments and statistical analysis show that the portable EMG acquisition system has significant differences in EMG signal characteristics between normal people and non-specific low back pain patients, and it has good measurement consistency and accuracy with the medical EMG acquisition equipment.

Core Technology of Wearable Multi-parameter Patient Monitor / 中国医疗器械杂志

Multi-parameters patient monitors are widely used in hospitals as medical device products, which have important clinical value. It expounds the core technologies of a miniature wearable multi-parameters patient monitor, and looks forward to its application prospects. In addition to traditional applications, when combined with a networked health service platforms, its applications will be greatly expanded in the context of big data and artificial intelligence technologies. The laboratory prototype of this product has been completed and has achieved the anticipative design goal.

Improving the Prediction of Protein-Protein Interaction Sites Using a Novel Over-Sampling Approach and Predicted Shape Strings.

Identification of protein-protein interaction (PPI) sites is one of the most challenging tasks in bioinformatics and many computational methods based on support vector machines have been developed. However, current methods often fail to predict PPI sites mainly because of the severe imbalance between the numbers of interface and non-interface residues. In this study, we propose a novel over-sampling method that relaxes the class-imbalance problem based on local density distributions. We applied the proposed method to a PPI dataset that includes 2,829 interface and 24,616 non-interface residues. The experimental result showed a significant improvement in predictive performance comparing with the other state-of-the-art methods according to the six evaluation measures.