Huey-Wen Liang’s research while affiliated with National Taiwan University Hospital and other places

Background Circadian rhythm disruptions are a common concern for poststroke patients undergoing rehabilitation and might negatively impact their functional outcomes. Objective Our research aimed to uncover unique patterns and disruptions specific to poststroke rehabilitation patients and identify potential differences in specific rest-activity rhythm indicators when compared to inpatient controls with non–brain-related lesions, such as patients with spinal cord injuries. Methods We obtained a 7-day recording with a wearable actigraphy device from 25 poststroke patients (n=9, 36% women; median age 56, IQR 46-71) and 25 age- and gender-matched inpatient control participants (n=15, 60% women; median age 57, IQR 46.5-68.5). To assess circadian rhythm, we used a nonparametric method to calculate key rest-activity rhythm indicators—relative amplitude, interdaily stability, and intradaily variability. Relative amplitude, quantifying rest-activity rhythm amplitude while considering daily variations and unbalanced amplitudes, was calculated as the ratio of the difference between the most active 10 continuous hours and the least active 5 continuous hours to the sum of these 10 and 5 continuous hours. We also examined the clinical correlations between rest-activity rhythm indicators and delirium screening tools, such as the 4 A’s Test and the Barthel Index, which assess delirium and activities of daily living. Results Patients who had a stroke had higher least active 5-hour values compared to the control group (median 4.29, IQR 2.88-6.49 vs median 1.84, IQR 0.67-4.34; P=.008). The most active 10-hour values showed no significant differences between the groups (stroke group: median 38.92, IQR 14.60-40.87; control group: median 31.18, IQR 18.02-46.84; P=.93). The stroke group presented a lower relative amplitude compared to the control group (median 0.74, IQR 0.57-0.85 vs median 0.88, IQR 0.71-0.96; P=.009). Further analysis revealed no significant differences in other rest-activity rhythm metrics between the two groups. Among the patients who had a stroke, a negative correlation was observed between the 4 A’s Test scores and relative amplitude (ρ=–0.41; P=.045). Across all participants, positive correlations emerged between the Barthel Index scores and both interdaily stability (ρ=0.34; P=.02) and the most active 10-hour value (ρ=0.42; P=.002). Conclusions This study highlights the relevance of circadian rhythm disruptions in poststroke rehabilitation and provides insights into potential diagnostic and prognostic implications for rest-activity rhythm indicators as digital biomarkers.

Background Computerized posturography obtained in standing conditions has been applied to classify fall risk for older adults or disease groups. Combining machine learning (ML) approaches is superior to traditional regression analysis for its ability to handle complex data regarding its characteristics of being high-dimensional, non-linear, and highly correlated. The study goal was to use ML algorithms to classify fall risks in community-dwelling older adults with the aid of an explainable artificial intelligence (XAI) approach to increase interpretability. Methods A total of 215 participants were included for analysis. The input information included personal metrics and posturographic parameters obtained from a tracker-based posturography of four standing postures. Two classification criteria were used: with a previous history of falls and the timed-up-and-go (TUG) test. We used three meta-heuristic methods for feature selection to handle the large numbers of parameters and improve efficacy, and the SHapley Additive exPlanations (SHAP) method was used to display the weights of the selected features on the model. Results The results showed that posturographic parameters could classify the participants with TUG scores higher or lower than 10 s but were less effective in classifying fall risk according to previous fall history. Feature selections improved the accuracy with the TUG as the classification label, and the Slime Mould Algorithm had the best performance (accuracy: 0.72 to 0.77, area under the curve: 0.80 to 0.90). In contrast, feature selection did not improve the model performance significantly with the previous fall history as a classification label. The SHAP values also helped to display the importance of different features in the model. Conclusion Posturographic parameters in standing can be used to classify fall risks with high accuracy based on the TUG scores in community-dwelling older adults. Using feature selection improves the model’s performance. The results highlight the potential utility of ML algorithms and XAI to provide guidance for developing more robust and accurate fall classification models. Trial registration Not applicable

BACKGROUND Circadian rhythm disruptions are a common concern for poststroke patients undergoing rehabilitation and might negatively impact their functional outcomes. OBJECTIVE Our research aimed to uncover unique patterns and disruptions specific to poststroke rehabilitation patients and identify potential differences in specific rest-activity rhythm indicators when compared to inpatient controls with non–brain-related lesions, such as patients with spinal cord injuries. METHODS We obtained a 7-day recording with a wearable actigraphy device from 25 poststroke patients (n=9, 36% women; median age 56, IQR 46-71) and 25 age- and gender-matched inpatient control participants (n=15, 60% women; median age 57, IQR 46.5-68.5). To assess circadian rhythm, we used a nonparametric method to calculate key rest-activity rhythm indicators—relative amplitude, interdaily stability, and intradaily variability. Relative amplitude, quantifying rest-activity rhythm amplitude while considering daily variations and unbalanced amplitudes, was calculated as the ratio of the difference between the most active 10 continuous hours and the least active 5 continuous hours to the sum of these 10 and 5 continuous hours. We also examined the clinical correlations between rest-activity rhythm indicators and delirium screening tools, such as the 4 A’s Test and the Barthel Index, which assess delirium and activities of daily living. RESULTS Patients who had a stroke had higher least active 5-hour values compared to the control group (median 4.29, IQR 2.88-6.49 vs median 1.84, IQR 0.67-4.34; P =.008). The most active 10-hour values showed no significant differences between the groups (stroke group: median 38.92, IQR 14.60-40.87; control group: median 31.18, IQR 18.02-46.84; P =.93). The stroke group presented a lower relative amplitude compared to the control group (median 0.74, IQR 0.57-0.85 vs median 0.88, IQR 0.71-0.96; P =.009). Further analysis revealed no significant differences in other rest-activity rhythm metrics between the two groups. Among the patients who had a stroke, a negative correlation was observed between the 4 A’s Test scores and relative amplitude (ρ=–0.41; P =.045). Across all participants, positive correlations emerged between the Barthel Index scores and both interdaily stability (ρ=0.34; P =.02) and the most active 10-hour value (ρ=0.42; P =.002). CONCLUSIONS This study highlights the relevance of circadian rhythm disruptions in poststroke rehabilitation and provides insights into potential diagnostic and prognostic implications for rest-activity rhythm indicators as digital biomarkers.

Postural control is reduced in patients with low back pain (LBP), which is considered an important factor attributing to the chronicity of LBP and a target for treatment. It is proposed that the changes in postural steadiness in sitting reflect the trunk control better than those in standing, but the previous study results are inconsistent. Thus, this study aimed to compare trunk sway parameters during standing and sitting between patients with chronic LPB (CLBP) and controls using a tracker-based posturography to record the trunk displacement trajectories at the lumbar level (TDL). A total of 64 participants (32 patients with CLBP and 32 pain-free controls) were included in this study. The postural sway was measured under four conditions, sitting or standing on unstable surface, with eyes open or closed. The TDL parameters were compared between the two groups to explore their discriminative ability. The CLBP group had more body sway than the control group, characterized by several TDL parameters in sitting with eyes closed and standing with eyes open. The TDL parameters with the highest area under the curve according to the receiver operating characteristic curve analysis were the root mean square distance and mean frequency in the medial-lateral direction obtained in the sitting with eyes closed. In conclusion, we confirmed the advantage of using sitting posturographic parameters as a sensitive measure to detect impaired trunk control in patients with CLBP. The results would help choose sensitive outcome measures to reflect the postural control of patients with LBP.

This paper investigates the applications of explainable AI (XAI) in healthcare, which aims to provide transparency, fairness, accuracy, generality, and comprehensibility to the results obtained from AI and ML algorithms in decision-making systems. The black box nature of AI and ML systems has remained a challenge in healthcare, and interpretable AI and ML techniques can potentially address this issue. Here we critically review previous studies related to the interpretability of ML and AI methods in medical systems. Descriptions of various types of XAI methods such as layer-wise relevance propagation (LRP), Uniform Manifold Approximation and Projection (UMAP), Local Interpretable Model-agnostic Explanations (LIME), SHapley Additive exPlanations (SHAP), ANCHOR, contextual importance and utility (CIU), Training calibration-based explainers (TraCE), Gradient-weighted Class Activation Mapping (Grad-CAM), t-distributed Stochastic Neighbor Embedding (t-SNE), NeuroXAI, Explainable Cumulative Fuzzy Class Membership Criterion (X-CFCMC) along with the diseases which can be explained through these methods are provided throughout the paper. The paper also discusses how AI and ML technologies can transform healthcare services. The usability and reliability of the presented methods are summarized, including studies on the usability and reliability of XGBoost for mediastinal cysts and tumors, a 3D brain tumor segmentation network, and the TraCE method for medical image analysis. Overall, this paper aims to contribute to the growing field of XAI in healthcare and provide insights for researchers, practitioners, and decision-makers in the healthcare industry. Finally, we discuss the performance of XAI methods applied in medical health care systems. It is also needed to mention that a brief implemented method is provided in the methodology section.

Background Reaction time (RT) is an important dependent variable to assess components of cognitive function. Thus, it has been a valuable parameter for research and clinical evaluation. A head-mounted display for virtual reality (HMD-VR) provides a standardized external visual condition and could be a promising tool for measuring RT. The purpose of this study is to establish the feasibility, reliability, and validity of HMD-VR-based software in evaluating simple RT (SRT). Methods Thirty healthy participants volunteered for the study. A SRT test was created by VIVE ProEye (HTC, Inc.). The participants responded with a keyboard when a square target was used in random intervals for 100 trials. To determine the optimal test repetition, the difference between the SRTS calculated with different trial numbers was analyzed. The one-week reliability of the median SRT was evaluated with the intraclass correlation coefficient (ICC). Finally, the convergent validity was tested by computing the correlation coefficient with a personal computer-based (PC-based) software, RehaComÒ (HASOMED, Inc.) with a similar task design. Results The median SRTs of the virtual reality-based (VR-based) and computer-based systems were 326.0 and 319.5 ms, respectively. Significantly longer RT obtained by the VR-based method was observed in the last 25-trial block for the non-dominant hand and bilateral hands according to Friedman’s test. The ICC was 0.71 (p<0.001), indicating good test-retest reliability. There was a high correlation (r=0.85~0.89) and good agreement between the VR-based and PC-based tests, with the VR-based SRT being 9-10 ms longer than the PC-based SRT according to Bland–Altman plots. Conclusions Our results supported the good reliability and high convergent validity of this HMD-VR-based RT testing. A test length of 50 trials was suggested to avoid possible decremental performance while maintaining good reliability. The program can be applied in future studies when spatial-specific RT is the main interest to provide a standardized external environment.

Coronavirus disease 2019 (COVID-19) has caused tremendous morbidity and mortality worldwide. The large number of post-COVID survivors has drawn attention to the management of post-COVID condition, known as long COVID. This review examines current knowledge of long COVID, regarding its epidemiology, mechanism, and clinical presentations in both adults and children. We also review the rehabilitation principles, modules, and effects, and share Taiwan's efforts to provide a top-down, nationwide care framework for long COVID patients. Dyspnea, chronic cough, and fatigue are the most commonly reported symptoms in the first 6 months after infection, but cognitive impairment and psychological symptoms may persist beyond this time. Several possible mechanisms behind these symptoms were proposed, but remained unconfirmed. These symptoms negatively impact individuals' function, activities, participation and quality of life. Rehabilitation is a key element of management to achieve functional improvement. Early management should start with comprehensive evaluation and identification of red flags. Exercise-based therapy, an essential part of management of long COVID, can be conducted with different modules, including telerehabilitation. Post-exertional symptom exacerbation and orthostatic hypotension should be carefully monitored during exercise. Randomized control trials with a large sample size are needed to determine the optimal timing, dosage, and modules.