Chun-Chuan Chen

BACKGROUND Breast cancer patients undergoing chemotherapy often experience high levels of stress and symptom burden. Mindful breathing virtual reality with biofeedback may offer an effective non-pharmacological intervention for stress management, yet its efficacy remains underexplored. OBJECTIVE This study evaluated the effect of mindful breathing virtual reality with biofeedback in reducing stress and symptom severity in breast cancer patients undergoing chemotherapy. METHODS A three-arm parallel randomized controlled trial was conducted from August to November 2023 at a medical center in Taiwan. A total of 82 participants were randomly assigned to the mindful breathing virtual reality group (n = 25), the mindful breathing audio training group (n = 28), and the control group (n = 29). Outcomes included stress, symptom severity, and physiological features such as electroencephalography, heart rate variability, respiratory signals, and galvanic skin response. Machine learning analyses were used to identify stress response patterns. RESULTS Participants in the mindful breathing virtual reality group showed a significant reduction in stress level (p < .05) and symptom severity (p < .001) compared to the other groups. Additionally, physiological features demonstrated improvements in electroencephalography, heart rate variability, respiratory signals, and galvanic skin response. Machine learning analyses further confirmed the superior accuracy of the mindful breathing virtual reality intervention in detecting and differentiating stress responses. CONCLUSIONS Mindful breathing virtual reality effectively reduces stress and symptom severity in breast cancer patients undergoing chemotherapy, with measurable physiological benefits. These findings highlight the potential of virtual reality as an innovative, non-pharmacological intervention to support stress management in oncology care. CLINICALTRIAL This study was registered on ClinicalTrials.gov under the identifier NCT06541587.

Methamphetamine use disorder (MUD) is a substance use disorder. Because MUD has become more prevalent due to the COVID-19 pandemic, alternative ways to help the efficiency of mass screening of MUD are important. Previous studies used electroencephalogram (EEG), heart rate variability (HRV), and galvanic skin response (GSR) aberrations during the virtual reality (VR) induction of drug craving to accurately separate patients with MUD from the healthy controls. However, whether these abnormalities present without induction of drug-cue reactivity to enable separation between patients and healthy subjects remains unclear. Here, we propose a clinically comparable intelligent system using the fusion of 5–channel EEG, HRV, and GSR data during resting state to aid in detecting MUD. Forty-six patients with MUD and 26 healthy controls were recruited and machine learning methods were employed to systematically compare the classification results of different fusion models. The analytic results revealed that the fusion of HRV and GSR features leads to the most accurate separation rate of 79%. The use of EEG, HRV, and GSR features provides more robust information, leading to relatively similar and enhanced accuracy across different classifiers. In conclusion, we demonstrated that a clinically applicable intelligent system using resting-state EEG, ECG, and GSR features without the induction of drug cue reactivity enhances the detection of MUD. This system is easy to implement in the clinical setting and can save a lot of time on setting up and experimenting while maintaining excellent accuracy to assist in mass screening of MUD.

The GO/NOGO task provides an objective assessment of a subject's attention and response inhibition and is typically given to subjects without any unexpected distractions. Studying the impact of distractions is important from the therapeutic viewpoint as distractions may occur during exposure therapy and degrade treatment efficacy. In this study, we utilized a virtual classroom integrated with electroencephalogram (EEG) for a GO/NOGO task with multimode environmental distractions to study the impact of distractions on behavioral and neuronal activities. Thirty healthy male adults were recruited. Statistical analysis and machine learning methods were employed to analyze the behavioral and neuronal data. The results demonstrated no significant behavioral differences between conditions with and without distractions. However, the impacts of distractions manifested in the enhancement of frequency-specific power, including theta, alpha, and gamma oscillations in GO trials, as well as beta power and the N200 peak in NOGO trials, highlighting their role in attention regulation and response inhibition. Finally, machine learning result analysis identified significant differences between conditions with and without distractions using EEG features, achieving an accuracy rate of 98.3%. In conclusion, we found that introducing distractions into a GO/NOGO task provides a deeper understanding of the neuronal correlates of distractions, and these findings can inform the development of therapeutic strategies for attention-related disorders.

Frailty, is becoming a more serious issue as the population ages. Numerous studies have shown that exercise can effectively slow the development of frailty. Compared with vigorous exercise, Baduanjin (BDJ), a kind of traditional Chinese qigong with eight simple movements, is more suitable for frailty patients. BDJ has been used to train frailty patients by physical therapists. To provide an enhanced training method, we designed a lightweight family-based frailty training system via a virtual BDJ coach. To achieve a compact system, we use a webcam as the main device. The system also supports the Kinect framework. We use pose estimation and motion recognition methods to analyze the user's movements. In addition, a novel transfer learning method is proposed. We designed a mapping model called “Skeleton Mapnet” to convert skeletal data from different frameworks. This method enables datasets from different frameworks to share classification models. It can also mix skeletal data from different frameworks to solve the lack of webcam datasets. Such a design allows the system to be easily ported into other platforms. In addition, the system is also suitable for the use of the Artificial Intelligence(AI) of Things. Our design ensures that frailty patients can easily learn and operate the system.

Depression not only afflicts hundreds of millions of people but also contributes to a global disability and healthcare burden. The primary method of diagnosing depression relies on the judgment of medical professionals in clinical interviews with patients, which is subjective and time-consuming. Recent studies have demonstrated that text, audio, facial attributes, heart rate, and eye movement could be utilized for depression-level stratification. In this paper, we construct a virtual assistant for automatic depression-level stratification on mobile devices that can actively guide users through voice dialogue and change conversation content using emotion perception. During the conversation, features from text, audio, facial attributes, heart rate, and eye movement are extracted for multi-modal depression-level stratification. We utilize a feature-level fusion framework to integrate five modalities and the deep neural network to classify the varying levels of depression, which include healthy, mild, moderate, or severe depression, as well as bipolar disorder (formerly called manic depression). With outcome data from 168 subjects, experimental results reveal that the total accuracy of feature-level fusion with five modal features achieves the highest accuracy of 90.26 percent.

Objective: Attention-deficit/hyperactivity disorder (ADHD) is a childhood-onset neurodevelopmental disorder with a prevalence ranging from 6.1 to 9.4%. The main symptoms of ADHD are inattention, hyperactivity, impulsivity, and even destructive behaviors that may have a long-term negative influence on learning performance or social relationships. Early diagnosis and treatment provide the best chance of reducing and managing symptoms. Currently, ADHD diagnosis relies on behavioral observations and ratings by clinicians and parents. Medical diagnosis of ADHD was reported to be delayed because of a global shortage of well-trained clinicians, the heterogeneous nature of ADHD, and combined comorbidities. Therefore, alternative ways to increase the efficiency of early diagnosis are needed. Previous studies used behavioral and neurophysiological data to assess patients with ADHD, yielding an accuracy range from 56.6% to 92%. Several factors were shown to affect the detection rate, including methods and tasks used and the number of electroencephalogram (EEG) channels. Given that children with ADHD have difficulty sustaining attention, in this study, we tested whether data from multiple tasks with different difficulties and prolonged experiment times can probe the levels of brain resources engaged during task performance and increase ADHD detection. Specifically, we proposed a Deep Neural Network-based (DNN) fusion model of multiple tasks to enhance the detection of ADHD. Methods & Results: Forty-nine children with ADHD and thirty-two typically developing children were recruited. Analytic results show that the fusion of multi-task neurophysiological data can increase the separation rate to 89%, whereas a single data type can only achieve a best accuracy of 81%. Moreover, the use of multiple tasks helps distinguish between children with ADHD and typically developing children. Our results suggest that different neurophysiological models from multiple tasks can provide essential information to assist in ADHD screening. In conclusion, the proposed model offers a more efficient, and accurate alternative for early clinical diagnosis and management of ADHD. The application of artificial intelligence and multimodal neurophysiological data in clinical settings sets a precedent for digital health, paving the way for future advancements in the field.

Human microbiota mainly resides on the skin and in the gut. Human gut microbiota can produce a variety of short chain fatty acids (SCFAs) that affect many physiological functions and most importantly modulate brain functions through the bidirectional gut-brain axis. Similarly, skin microorganisms also have identical metabolites of SCFAs reported to be involved in maintaining skin homeostasis. However, it remains unclear whether these SCFAs produced by skin bacteria can affect brain cognitive functions. In this study, we hypothesize that the brain’s functional activities are associated with the skin bacterial population and examine the influence of local skin-bacterial growth on event-related potentials (ERPs) during an oddball task using EEG. Additionally, five machine learning (ML) methods were employed to discern the relationship between skin microbiota and cognitive functions. Twenty healthy subjects underwent three rounds of tests under different conditions—alcohol, glycerol, and water. Statistical tests confirmed a significant increase in bacterial population under water and glycerol conditions when compared to the alcohol condition. The metabolites of bacteria can turn phenol red from red–orange to yellow, confirming an increase in acidity. P3 amplitudes were significantly enhanced in response to only oddball stimulus at four channels (Fz, FCz, and Cz) and were observed after the removal of bacteria when compared with that under the water and glycerol manipulations. By using machine learning methods, we demonstrated that EEG features could be separated with a good accuracy (> 88%) after experimental manipulations. Our results suggest a relationship between skin microbiota and brain functions. We hope our findings motivate further study into the underlying mechanism. Ultimately, an understanding of the relationship between skin microbiota and brain functions can contribute to the treatment and intervention of diseases that link with this pathway.

Biosynthesis of gamma-aminobutyric acid (GABA) can be achieved by naturally occurring microorganisms with the advantages of cost-effectiveness and safety. In this study, Bacillus amyloliquefaciens EH-9 strain (B. amyloliquefaciens EH-9), a soil bacterium, was used to promote the accumulation of GABA in germinated rice seed. Further, the topical application of supernatant from rice seed co-cultivated with soil B. amyloliquefaciens EH-9 can significantly increase the production of type I collagen (COL1) in the dorsal skin of mice. The knocking down of the GABA-A receptor (GABA_A) significantly reduced the production of COL1 in the NIH/3T3 cells and in the dorsal skin of mice. This result suggests that topical application of GABA can promote the biosynthesis of COL1 via its interaction with the GABA_A receptor in the dorsal skin of mice. In summary, our findings illustrate for the first time that soil B. amyloliquefaciens EH-9 elicits GABA production in germinated rice seed to upregulate the formation of COL1 in the dorsal skin of mice. This study is translational because the result shows a potential remedy for skin aging by stimulating COL1 synthesis using biosynthetic GABA associated with B. amyloliquefaciens EH-9.