Meng-Hsiun Tsai

Background/Objectives: Extracranial–intracranial (EC-IC) bypass, particularly superficial temporal artery to middle cerebral artery (STA-MCA) anastomosis, is used to augment cerebral perfusion in selected patients with moyamoya disease or atherosclerotic steno-occlusive disease. This study evaluated the impact of STA-MCA bypass on regional cerebral perfusion using a semi-quantitative SPECT area-based method. Methods: Of 40 patients screened between June 2018 and June 2022, 15 with complete paired pre- and postoperative CTA and SPECT data were included. Postoperative imaging was performed at a median of 12 months (range, 6–15 months). Proportional perfusion area in the basal ganglia and temporal regions was calculated using Labelme annotation, Canny edge detection, and Green’s theorem, normalized to total brain area. Paired comparisons and exploratory subgroup analyses were performed. Normality was assessed by Shapiro–Wilk test; Wilcoxon signed-rank tests served as sensitivity analyses. Results: Mean age was 56 ± 15.7 years (40% female). All 16 anastomoses were patent. Postoperative perfusion area increased significantly in the basal ganglia (mean difference 0.24, 95% CI 0.13–0.35, p = 0.0003) and temporal regions (0.34, 95% CI 0.08–0.60, p = 0.012). Wilcoxon tests confirmed these findings (basal ganglia p = 0.0003; temporal p = 0.008). In unilateral cases, ipsilateral and contralateral basal ganglia improvements did not differ significantly (p = 0.839). Results should be interpreted as exploratory given the limited sample size. Conclusions: STA-MCA bypass was associated with significant increases in semi-quantitative perfusion area on SPECT, particularly in the basal ganglia. The lack of ipsilateral–contralateral difference is consistent with a possible global hemodynamic effect, but this interpretation remains hypothesis-generating. The pragmatic SPECT method is feasible for perioperative monitoring yet requires further validation. Larger prospective studies with standardized protocols are warranted.

Background/Objectives: Sepsis is clinically heterogeneous, and comorbidities may alter physiological patterns associated with sepsis status. This study examined whether comorbidity-specific modeling revealed differences in retrospective sepsis discrimination. Methods: MIMIC-IV data were analyzed across 27 Elixhauser-based analytical subgroups using Random Forest, XGBoost, CatBoost, LightGBM, AdaBoost, multilayer perceptron, TabNet, and FT-Transformer. Available neural-model scripts used a stratified 80/20 training–evaluation split. Class imbalance was addressed using training-partition SMOTE and model-specific loss or sampling settings. SHAP was used to describe model-attributed feature contributions. Results: Tree-based ensembles achieved ROC-AUC values above 0.80 in many subgroups, whereas neural models showed more variable performance and trade-offs between discrimination and sepsis-class recall. Red cell distribution width, age, and magnesium were frequently emphasized by the fitted models. Conclusions: The findings demonstrate subgroup- and model-specific differences in retrospective sepsis-status discrimination. Because a verified pre-sepsis feature-time boundary was not enforced, the results should not be interpreted as prospectively validated early prediction. External validation, calibration, and decision-analytic evaluation are required before clinical use.

Background/Objectives: Extracranial–Intracranial (EC-IC) arterial bypass, particularly superficial temporal artery to middle cerebral artery (STA-MCA) anastomosis, is employed in selected patients with moyamoya disease or atherosclerotic steno-occlusive disease to augment cerebral perfusion. Objective hemodynamic monitoring before and after revascularization remains essential. This study evaluated the impact of STA-MCA bypass on regional cerebral blood flow (rCBF) using single-photon emission computed tomography (SPECT) with a semi-quantitative area-based method. The proportional perfusion area was used as a pragmatic surrogate marker of regional cerebral blood flow (rCBF) changes. Methods: Fifteen patients who underwent STA-MCA bypass between June 2018 and June 2022 with available pre- and postoperative CTA and rCBF-SPECT studies were retrospectively analyzed. During the study period, 40 patients with complete paired pre- and postoperative CTA and SPECT imaging were included after exclusion of those with missing data or loss to follow-up. Preoperative and postoperative clinical evaluations were performed by the attending neurosurgical team. Postoperative CTA and rCBF-SPECT were typically obtained between 6 and 15 months after surgery, with timing individualized according to clinical status. A custom semi-quantitative technique using Labelme annotation, Canny edge detection, and Green’s theorem was applied to calculate the proportional perfusion area in basal ganglia and temporal regions, normalized to total brain area. Paired pre–post comparisons and subgroup analyses (left- vs. right-sided surgery; ipsilateral vs. contralateral regions) were performed. Normality of differences was assessed with the Shapiro–Wilk test. Post hoc power analyses were conducted. Sensitivity analyses used the Wilcoxon signed-rank test. Subgroup analyses were considered exploratory. Results: Mean age was 56 ± 15.7 years (40% female). Indications included moyamoya disease (20%) and arterial stenosis/occlusion (80%). All 16 anastomoses (14 unilateral, 1 bilateral) were patent postoperatively. One patient underwent bilateral bypass; this patient was included in overall analyses but excluded from ipsilateral/contralateral comparisons to allow clear laterality classification. Overall, postoperative perfusion area proportion improved significantly in basal ganglia (mean difference 0.24, 95% CI 0.13–0.35, p = 0.0003; post hoc power 0.993) and temporal regions (0.34, 95% CI 0.08–0.60, p = 0.013; post hoc power 0.76). Normality assumptions were satisfied (Shapiro–Wilk p > 0.05); Wilcoxon signed-rank tests confirmed significance (basal ganglia p = 0.0003; temporal p = 0.008). Improvement in basal ganglia perfusion was similar between left-sided (Δ 0.23) and right-sided (Δ 0.28) surgery groups. In unilateral cases, there was no significant difference between ipsilateral and contralateral basal ganglia improvement (p = 0.839), consistent with the hypothesis of a more global hemodynamic effect rather than a strictly localized phenomenon. These findings should be interpreted as exploratory given the limited sample size. Conclusions: In our study, STA-MCA bypass was associated with a significant increase in semi-quantitative perfusion area on SPECT, with consistent benefits observed in the basal ganglia regardless of surgical laterality. The absence of a significant ipsilateral–contralateral difference supports the hypothesis that hemodynamic benefit may extend beyond the immediate surgical territory, possibly mediated by increased overall inflow, reversal of steal phenomenon, and collateral recruitment. The pragmatic semi-quantitative SPECT method is feasible for perioperative hemodynamic monitoring but requires further validation against quantitative perfusion techniques. Larger prospective multicenter studies incorporating quantitative imaging and standardized neuropsychological assessment are warranted to confirm these observations and refine patient selection.

Tennis match analysis is a critical component of sports science, offering data on player performance, workload management, and competitive stress. We developed a data-driven framework to classify tennis matches as high-stress or low-stress using the Association of Tennis Professionals’ match statistics. High-stress matches are characterized by extended duration or frequent break points, both representing elevated physical and psychological demands. We implement TabNet and compare its performance with recurrent deep learning models, including long short-term memory (LSTM), bidirectional LSTM, attention-enhanced LSTM, and convolutional LSTM. Experimental results show that TabNet achieves the best accuracy (98%), while the recurrent models maintain accuracies above 93%, demonstrating consistent predictive capability. To enhance interpretability, SHAP analysis identifies break points faced, break points saved, and match duration as the most influential determinants of match stress, with serving and returning features providing secondary contributions. These findings confirm the effectiveness of interpretable deep learning in sports analytics and highlight its potential for guiding training and match preparation.

Tumor-educated platelets (TEPs) have emerged as promising biomarkers for liquid biopsy-4based cancer diagnostics. However, the intrinsic challenges of transcriptomic data, including high dimensionality, small sample size, and strong nonlinearity, hinder robust classification. In this study, we develop a deep learning model that integrates long short-term memory (LSTM), bidirectional LSTM, convolutional neural network–LSTM (CNN-LSTM), and attention-based LSTM (Attention-LSTM) to classify cancer and healthy samples from the GSE68086 dataset (235 cancer, 50 healthy). Experimental results demonstrate that all models achieve competitive performance, with overall accuracies approaching 0.88, while the Attention-LSTM and CNN-LSTM further enhance interpretability and local feature extraction. Moreover, SHAP analysis confirms that selected genes contributing to classification align with biologically meaningful signals, providing both predictive accuracy and clinical interpretability. These findings highlight the potential of deep sequential models in advancing platelet-based cancer diagnostics.

Background: Electronic patient-reported outcomes (ePROs) enable real-time symptom monitoring and early intervention in oncology. Large language models (LLMs), when combined with retrieval-augmented generation (RAG), offer scalable Artificial Intelligence (AI)-driven education tailored to individual patient needs. However, few studies have examined the feasibility and clinical impact of integrating ePRO with LLM-RAG feedback during radiotherapy in high-toxicity settings such as head and neck cancer. Methods: This prospective observational study enrolled 42 patients with head and neck cancer undergoing radiotherapy from January to December 2024. Patients completed ePRO entries twice weekly using a web-based platform. Following each entry, an LLM-RAG system (Gemini 1.5-based) generated real-time educational feedback using National Comprehensive Cancer Network (NCCN) guidelines and institutional resources. Primary outcomes included percentage weight loss and treatment interruption days. Statistical analyses included t-tests, linear regression, and receiver operating characteristic (ROC) analysis. A threshold of ≥6 ePRO entries was used for subgroup analysis. Results: Patients had a mean age of 53.6 years and submitted an average of 8.0 ePRO entries. Frequent ePRO users (≥6 entries) had significantly less weight loss (4.45% vs. 7.57%, p = 0.021) and fewer treatment interruptions (0.67 vs. 2.50 days, p = 0.002). Chemotherapy, moderate-to-severe pain, and lower ePRO submission frequency were associated with greater weight loss. ePRO submission frequency was negatively correlated with both weight loss and treatment interruption days. The most commonly reported symptoms were appetite loss, fatigue, and nausea. Conclusions: Integrating LLM-RAG feedback with ePRO systems is feasible and may enhance symptom control, treatment continuity, and patient engagement in head and neck cancer radiotherapy. Further studies are warranted to validate the clinical benefits of AI-supported ePRO platforms in routine care.

This study aims to evaluate whether safety nets on construction sites are correctly installed using an image processing and deep learning technique. The developed method performs data preprocessing, including horizontal flipping, rotation, and contrast-limited adaptive histogram equalization, and then applies the YOLO model to estimate the accuracy of safety net installation. The developed method significantly improved the accuracy of the YOLO model detection and mitigated errors associated with large safety net surfaces and slanted steel beams using the novel linear intersection over union as a metric. The proposed method effectively improved the assessment of safety net installation.

Glioblastoma multiforme (GBM) is the most aggressive and common primary brain tumor. Magnetic resonance imaging (MRI) provides detailed visualization of tumor morphology, edema, and necrosis. However, manually segmenting GBM from MRI scans is time-consuming, subjective, and prone to inter-observer variability. Therefore, automated and reliable segmentation methods are crucial for improving diagnostic accuracy. This study employs an image semantic segmentation model to segment brain tumors in MRI scans of GBM patients. The MRI recall images include T1-weighted imaging (T1WI) and fluid-attenuated inversion recovery (FLAIR) sequences. To enhance the performance of the semantic segmentation model, image preprocessing techniques were applied before analyzing and comparing commonly used segmentation models. Additionally, a survival model was constructed using discrete genotype attributes of GBM patients. The results indicate that the DeepLabV3+ model achieved the highest accuracy for semantic segmentation, with an accuracy of 77.9% on T1WI image sequences, while the U-Net model achieved 80.1% accuracy on FLAIR image sequences. Furthermore, in constructing the survival model using a discrete attribute dataset, the dataset was divided into three subsets based on different missing value handling strategies. This study found that replacing missing values with 1 resulted in the highest accuracy, with the Bernoulli Bayesian model and the multinomial Bayesian model achieving an accuracy of 94.74%. This study integrates image preprocessing techniques and semantic segmentation models to improve the accuracy and efficiency of brain tumor segmentation while also developing a highly accurate survival model. The findings aim to assist physicians in saving time and facilitating preliminary diagnosis and analysis.

MiRNAs (microRNAs) are endogenous RNAs with lengths of 18 to 24 nucleotides and play critical roles in gene regulation and disease progression. Although traditional wet-lab experiments provide direct evidence for miRNA-disease associations, they are often time-consuming and complicated to analyze by current bioinformatics tools. In recent years, machine learning (ML) and deep learning (DL) techniques are powerful tools to analyze large-scale biological data. Hence, developing a model to predict, identify, and rank connections in miRNAs and diseases can significantly enhance the precision and efficiency in investigating the relationships between miRNAs and diseases. In this study, we utilized miRNA-disease association data obtained by biotechnological experiments to develop a DL model for miRNA-disease associations. To improve the accuracy of prediction in this model, we introduced two labeling strategies, weight-based and majority-based definitions, to classify miRNA-disease associations. After preprocessing, data was trained with a novel model combining gated recurrent units (GRU) and graph convolutional network (GCN) to predict the level of miRNA-disease associations. The miRNA-disease association datasets were from HMDD (the Human miRNA Disease Database) and categorized by two distinct labeling approaches, weight-based definitions and majority-based definitions. We classified the miRNA-disease associations into three groups, “upregulated”, “downregulated” and “nonspecific”, by regression analysis and multiclass classification. This GRU-GCN coordinated model achieved a robust area under the curve (AUC) score of 0.8 in all datasets, demonstrating the efficacy in predicting potential miRNA-disease relationships. By introducing innovative label-preprocessing methods, this study addressed the relationships between miRNAs and diseases, and improved the ambiguity of the results in different experiments. Based on these refined label definitions, we developed a DL-based model to refine and predict the results of associations between miRNAs and diseases. This model offers a valuable tool for complementing traditional experimental methods and enhancing our understanding of miRNA-related disease mechanisms.