Tag: Taiwan

In busy emergency departments, every minute and every medical resource matters. An AI-driven prediction model developed by Taipei Medical University may help physicians determine more efficiently whether patients need CT scans, supporting faster decision-making, reducing unnecessary imaging, and improving hospital resource use.

The model learns from real-world emergency department records, including mixed languages notes, abbreviations, and fragmented symptom descriptions. By turning unstructured medical language into practical decision support, the system demonstrates how artificial intelligence can assist frontline clinicians in complex and fast-paced care settings.

The study was led by Professor Yung-Chun Chang from the Graduate Institute of Data Science at Taipei Medical University (TMU), in collaboration with Dr. Ting-Yun Huang from the Emergency Department of Shuang-Ho Hospital.

Addressing Emergency Department Overcrowding

Emergency department overcrowding remains a critical global healthcare challenge affecting treatment efficiency, patient flow, and medical safety. Computed tomography (CT) imaging is essential for diagnosing serious and potentially life-threatening conditions, but previous studies have estimated that 20–40% of CT scans may be unnecessary. Overuse of CT imaging can increase healthcare costs, place pressure on hospital resources, and expose patients to avoidable radiation.

To address this issue, the research team developed an AI-based predictive system designed to support early clinical decision-making and optimize medical resource allocation.

Turning Clinical Language into AI Decision Support

A key innovation of this study lies in its ability to tackle the complexities of real-world emergency departments. Clinical records in Taiwanese hospitals often include mixed Chinese–English terminology, domain-specific abbreviations, and institution-specific expressions. In emergency settings, these notes are often brief, fragmented, and highly unstructured due to time constraints and urgent clinical workflows.

To overcome these challenges, the research team developed a comprehensive clinical language engineering pipeline capable of processing multilingual and highly heterogeneous medical text. The system performs automated language normalization, terminology standardization, correction of inconsistent clinical expressions, and semantic representation learning from noisy narrative data

This enables the model to extract meaningful diagnostic information from incomplete or irregular documentation and improves its robustness in real-world clinical settings.

The proposed framework integrates large-scale clinical text processing, multimodal feature fusion, prompt-based feature representation, and data augmentation using large language models. Unlike conventional prediction methods that rely mainly on structured numerical data or laboratory results, the system leverages unstructured clinical narrative text—such as chief complaints, symptom descriptions, medical history, and pain severity—as the primary source of predictive information. This design allows the model to capture subtle clinical patterns and contextual relationships often overlooked by traditional machine learning approaches.

Using 165,391 emergency department records from Shuang-Ho Hospital, the model demonstrated strong predictive performance, achieving an AUROC of 0.88. It showed high reliability in identifying patients who were unlikely to require CT imaging and outperformed traditional machine learning methods as well as existing biomedical language models.

The system was also developed with practical clinical implementation in mind. After training, the model can operate on standard hospital computing infrastructure and has the potential to be integrated with electronic health record systems, enabling real-time clinical decision support in diverse healthcare environments.

Advancing AI-Enabled Healthcare Innovation

This research highlights the importance of interdisciplinary collaboration between AI engineers and frontline clinicians. It also demonstrates the transformative potential of large language models in healthcare system optimization.

By addressing the complexities of multilingual and unstructured clinical data, the study provides a foundation for future AI-driven medical decision support systems. The research was supported by Taiwan’s National Science and Technology Council and published in Engineering Applications of Artificial Intelligence, an international journal with a 2024 impact factor of 8.0 and a ranking in the top 2.5% of the Engineering, Multidisciplinary category. These achievements reflect TMU’s growing impact in AI-enabled healthcare innovation and its commitment to advancing precision medicine and intelligent healthcare.

Look for More Information

• Original Article: Enhancing the Effectiveness of Emergency Department Computed Tomography Scans Using Pre-Trained Language Models
• Author Profile: Yung-Chun Chang, Professor, Graduate Institute of Data Science, College of Management

Major depressive disorder (MDD) remains one of the world’s most disabling conditions, yet finding an effective antidepressant treatment often still relies on a prolong trial-and-error process. A research team lead by Professor Syu-Jyun Peng has developed an EEG-based, machine-learning drug-response prediction system designed to help clinicians identify whether a patient is likely to respond to medication within the first weeks of treatment.

The project, “AI EEG Drug-Response Prediction System: Longitudinal Tracking and Personalized Recommendations for Depression Treatment,” received the 2025 IIP International Inventor Prize – International Innovation and Invention Elite Award, highlighting its potential contribution to precision psychiatry and personalized mental healthcare.

Turning Early EEG Signals into Actionable Predictions

Although antidepressants are widely used, patients often need to try multiple medications before finding an effective treatment. This delay can prolong symptoms, increase relapse risk, and place additional burden on patients, families, and healthcare systems.

Associate Professor Peng’s team developed the system based on a key insight: brain network activity may begin to change soon after medication starts, before clinical improvement becomes clearly visible. By capturing these early neurophysiological changes through electroencephalography, or EEG, the system aims to provide objective biomarkers that can support earlier and more individualized treatment decisions.

The study collected EEG recordings from 77 patients with major depressive disorder at baseline (Week 0) and after one week of medication (Week 1). These data were then used to predict treatment response at Weeks 4, 6, and 8 through machine-learning models combining EEG-derived features and clinical variables.

A Practical Pipeline for Real-World Clinics

The system was designed with real-world clinical application in mind. EEG is non-invasive, relatively affordable, and suitable for repeated monitoring, making it a promising tool for psychiatric care.

In the research pipeline, EEG signals are preprocessed and divided into canonical frequency bands (delta, theta, alpha, and beta) and converted into interpretable features.

The team focused on two complementary domains:

1. Power-based metrics (absolute and relative power), capturing spectral characteristics of brain activity.
2. Functional connectivity and phase synchronization metrics—including PLV, PLI, and wPLI, which measure how different brain regions coordinate with one another.

The team also computed a change index, comparing Week 1 and Week 0 EEG patterns, to identify early brain changes that may predict later treatment outcomes.

Promising Prediction Performance

The system achieved strong predictive results, with accuracy reaching 83.1% for Week 4, 73.3% for Week 6, and 80.0% for Week 8 treatment response.

Functional connectivity and phase synchronization features consistently emerged as major contributors to predictive performance. These findings support the idea that early changes in brain network dynamics may serve as meaningful biomarkers for antidepressant response.

From Research Prototype to Precision Psychiatry

The “heart” of this project was never just algorithm tuning—it was learning to build something that clinicians can confidently trust. Early on, EEG data can be complex and affected by artifacts, individual differences, and recording conditions. To address these challenges, the team refined preprocessing methods, strengthened artifact handling, and selected interpretable features with clinical relevance.

Following recognition at the 2025 IIP International Inventor Prize, the project will move toward larger-scale validation and clinical translation. Future development will focus on multi-site studies, clinician-friendly decision-support interface, multimodal data integration, and broader healthcare applications.

By helping clinicians identify likely treatment responders earlier, the system has the potential to reduce ineffective medication trials, shorten time to remission, and support more personalized depression care.

Advancing Personalized Mental Healthcare

This award-winning AI EEG drug-response prediction system represents an important step toward precision psychiatry. By combining EEG, machine learning, and longitudinal treatment tracking, Associate Professor Peng’s team aims to transform early brain responses into actionable clinical insights, which is helping patients move more quickly toward effective care.

Look for More Information

Original Article: Predicting the longitudinal efficacy of medication for depression using electroencephalography and machine learning

Keywords: Depression; Electroencephalography; Functional connectivity; Machine learning

Author Profile: Syu-Jyun Peng, Professor, In-Service Master Program in Artificial Intelligence in Medicine, College of Medicine

What if one of the most fatal cardiovascular emergencies could be stopped before it even begins?

A research team from Taipei Medical University (TMU) is redefining the future of cardiovascular care. In a breakthrough study published in the Journal of Nanobiotechnology, TMU researchers have introduced a new concept in cardiovascular medicine – “Biological Structural Intervention,” a pathology-tailored nanotherapy that targets the cellular roots of aortic weakening, shifting care from passive “watchful waiting” to a proactive prevention of aortic dissection (AD).

Ending the Era of “Watchful Waiting”

Aortic dissection is a life-threatening emergency where the inner layer of the aorta tears, often leading to rapid deterioration or sudden death. Currently, patients at risk are managed through hemodynamic control—lowering blood pressure to reduce stress on the wall. However, this “hemodynamic holding” does not address the underlying biological degradation, leaving many patients at risk of late-stage complications.

“Our goal was to move beyond simply managing pressure,” says Professor Chun-Che Shih, Vice Superintendent of TMU Wan-Fang Hospital. “We have developed a way to actively intervene in the biological structure of the aorta, reinforcing it at the cellular level before a rupture happens”.

The Innovation: A Biological Shield

The TMU team developed “triple-responsive” nanoparticles (MPCR NPs) that act as a precision biological shield. Rather than circulating broadly like traditional systemic drugs, these nanoparticles target Galectin-3 (Gal-3)—a protein that serves as a persistent “homing beacon” for inflammation and structural weakening.

Key features of this Biological Structural Intervention include:

• Precision Targeting: The nanoparticles achieve a selective accumulation in diseased aortic tissue, ensuring therapeutic action is concentrated exactly where the wall is failing.
• Triple-Responsive Activation: The therapy remains dormant until it senses the specific acidic pH, enzymatic activity, and oxidative stress found at the site of aortic damage.
• Multimodal Repair: Once activated, the system releases a combination of Nitric Oxide (NO) and Resveratrol (RES) to stabilize vascular muscle cells, restore the protective vessel lining, and block the enzymatic destruction of the aortic wall.

Transforming the Future of Cardiovascular Care

This research establishes a new paradigm: Active Prevention. By reinforcing the aorta’s structural integrity biologically, this technology bridges the dangerous clinical gap between daily blood pressure pills and invasive emergency surgery.

“This platform demonstrates how nanomedicine can move from simply delivering drugs to actively shaping the biological environment,” explains Professor Fwu-Long Mi. “It provides a scalable strategy for the preemptive treatment of complex vascular disorders”.

The findings mark a significant step toward a future where “precision prevention” replaces “emergency reaction,” potentially saving lives by ensuring the aorta never reaches a breaking point.

Look for More Information

Original Article:

Pathology-tailored nanotherapy via Galectin-3-targeted and triple-responsive nanoparticles enables multimodal therapy against aortic dissection

As high-temperature cooking methods such as roasting, grilling, and frying become increasingly common in modern diets, food scientists are paying closer attention to heat-induced contaminants that may pose health risk. While glycidyl esters (GEs) and 3-monochloropropanediol esters (3-MCPDEs) have long been associated with refined edible oils, emerging research suggests these compounds may also form directly in meat during cooking. To better understand how cooking conditions affect contaminant formation, researchers from Taipei Medical University (TMU) and the University of California, Davis (UC Davis) conducted a pioneering study investigating the relationship between heating conditions, fat composition, and lipid oxidation in meat products.

A research team led by Associate Professor Wei-Ju Lee from the School of Food Safety, College of Nutrition at TMU, in collaboration with Professor Selina C. Wang of UC Davis, has published new findings on the formation of heat-induced contaminants in meat. Their study, “Effects of oven heating on the formation of glycidyl esters and 3-monochloropropanediol esters in various meats”, has been published in the international journal Food Chemistry.

Previous studies have shown that GEs and 3-MCPDE are commonly found in refined vegetable oils and processed foods containing such oils. These compounds are considered potential carcinogenic contaminants and are primarily formed during the high-temperature deodorization stage of edible oil refining. In recent years, researchers have also discovered that endogenous fats in foods may generate GEs and 3-MCPDEs during cooking and thermal processing. Meat products, in particular, have attracted growing attention; however, the mechanism underlying the formation of these contaminants and the factors influencing their production have remained unclear.

This study is the first to establish quantitative relationships among fat composition, lipid oxidation, and contaminant formation under different heating conditions using real meat products as research samples. By employing actual meat specimens, the study overcomes the limitations of previous studies that focused on vegetable oil models and fills gaps in the literature. The research examined four commonly consumed meats with varying fat contents (pork loin, pork belly, beef belly, and chicken thigh), which were heated in a high-temperature oven to evaluate the effects of different cooking temperatures (150–300°C) and durations (10–30 minutes) on the formation of GEs and 3-MCPDEs.

The results demonstrated that contaminant concentration increased with both heating temperature and cooking duration, reaching peak levels after heating at 300°C for 30 minutes. At the same time, the meat samples experienced substantial moisture loss. The researchers also found positive correlations between fat content, lipid oxidation indicators, and the concentrations of GEs and 3-MCPDEs. In high-fat meat products, the two contaminants were also highly correlated.

These findings confirm that fat content and lipid oxidation are key factors promoting GEs and 3-MCPDEs during the thermal processing of meat products. The study provides important scientific evidence for improving meat preparation safety and offers valuable insights for the development of safer cooking practices.

Look for More Information

• Original Article: Effects of oven heating on the formation of glycidyl esters and 3-monochloropropanediol esters in various meats
• Keywords: Moisture content, Fat content, Fatty acid composition, Acid value, Total polar compound, Totox value
• Author Profile: Wei-Ju Lee, Associate Professor, School of Food Safety, College of Nutrition