A Deep-Ensemble-Level-Based Interpretable Takagi-Sugeno-Kang Fuzzy Classifier for Imbalanced Data

A Novel AUC Maximization Imbalanced Learning Approach for Predicting Composite Outcomes in COVID-19 Hospitalized Patients

The COVID-19 patient data for composite outcome prediction often comes with class imbalance issues, i.e., only a small group of patients develop severe composite events after hospital admission, while the rest do not. An ideal COVID-19 composite outcome prediction model should possess strong imbalanced learning capability. The model also should have fewer tuning hyperparameters to ensure good usability and exhibit potential for fast incremental learning. Towards this goal, this study proposes a novel imbalanced learning approach called Imbalanced maximizing-Area Under the Curve (AUC) Proximal Support Vector Machine (ImAUC-PSVM) by the means of classical PSVM to predict the composite outcomes of hospitalized COVID-19 patients within 30 days of hospitalization. ImAUC-PSVM offers the following merits: (1) it incorporates straightforward AUC maximization into the objective function, resulting in fewer parameters to tune. This makes it suitable for handling imbalanced COVID-19 data with a simplified training process. (2) Theoretical derivations reveal that ImAUC-PSVM has the same analytical solution form as PSVM, thus inheriting the advantages of PSVM for handling incremental COVID-19 cases through fast incremental updating. We built and internally and externally validated our proposed classifier using real COVID-19 patient data obtained from three separate sites of Mayo Clinic in the United States. Additionally, we validated it on public datasets using various performance metrics. Experimental results demonstrate that ImAUC-PSVM outperforms other methods in most cases, showcasing its potential to assist clinicians in triaging COVID-19 patients at an early stage in hospital settings, as well as in other prediction applications.

Closed-Loop Phase-Dependent Vibration Stimulation Improves Motor Imagery-Based Brain-Computer Interface Performance

The motor imagery (MI) paradigm has been wildly used in brain-computer interface (BCI), but the difficulties in performing imagery tasks limit its application. Mechanical vibration stimulus has been increasingly used to enhance the MI performance, but its improvement consistence is still under debate. To develop more effective vibration stimulus methods for consistently enhancing MI, this study proposes an EEG phase-dependent closed-loop mechanical vibration stimulation method. The subject's index finger of the non-dominant hand was given 4 different vibration stimulation conditions (i.e., continuous open-loop vibration stimulus, two different phase-dependent closed-loop vibration stimuli and no stimulus) when performing two tasks of imagining movement and rest of the index finger from his/her dominant hand. We compared MI performance and brain oscillatory patterns under different conditions to verify the effectiveness of this method. The subjects performed 80 trials of each type in a random order, and the average phase-lock value of closed-loop stimulus conditions was 0.71. It was found that the closed-loop vibration stimulus applied in the falling phase helped the subjects to produce stronger event-related desynchronization (ERD) and sustain longer. Moreover, the classification accuracy was improved by about 9% compared with MI without any vibration stimulation (p = 0.012, paired t-test). This method helps to modulate the mu rhythm and make subjects more concentrated on the imagery and without negative enhancement during rest tasks, ultimately improves MI-based BCI performance. Participants reported that the tactile fatigue under closed-loop stimulation conditions was significantly less than continuous stimulation. This novel method is an improvement to the traditional vibration stimulation enhancement research and helps to make stimulation more precise and efficient.