National Yang Ming Chiao Tung University

05/02/2024 | Press release | Distributed by Public on 05/01/2024 21:45

Advancing Precision Healthcare: NYCU Utilizes AI and Magneto-Electric Stimulation Techniques to Develop Diagnostic and Therapeutic Tools


Translated by Hsuchuan
Edited by Chance Lai
______
On April 16th, at the 2024 Biotechnology Research and Development Achievement Matchmaking Conference hosted by the National Health Research Institutes, National Yang Ming Chiao Tung University (NYCU), and China Medical University, 12 academic research teams presented their research achievements.

Among them, three teams from NYCU focused on precision detection/treatment, utilizing AI, magneto-electric stimulation, and other technologies to develop multiple precision medical tools.

AI Medical Imaging Stability Inspection Platform: Predicting Model Application Issues in Advance across Different Contexts

The team led by Professor You-Yin Chen and Dr. Ching-Fu Wang from the Department of Biomedical Engineering at NYCU announced the successful launch of their "Artificial Intelligence/Machine Learning Medical Imaging Inspection Platform." Assessing the stability of AI medical imaging models can eliminate model uncertainty and reduce inspection concerns.

By incorporating interference factors through generative techniques, researchers can predict in advance during the early development stages whether models will encounter issues when applied in different scenarios in the future. For instance, adaptability in various hospital environments.

Dr. Ching-Fu Wang pointed out that the platform currently supports various applications, including general imaging AI systems, systems combining 1D physiological signals with 2D imaging AI technology, and medical imaging systems. However, due to the current lack of corresponding regulatory standards, they also hope that this achievement will serve as a reference for future regulations.

Health Risk Prediction Technology: Providing Strong Support for Early Disease Prediction and Risk Assessment

In addition, the team led by Dr. Vincent S. Tseng from the Department of Computer Science has also brought cutting-edge health risk prediction technology. They applied deep learning and multi-objective optimization techniques to develop a series of technologies capable of "early anomaly prediction" (Snippet Policy Network) for physiological signals. In addition, with multimodal learning techniques, they integrated various data types, providing reliable assistance to physicians in diagnosis.

Doctoral student Wei-Yun Hsu from the team stated that their developed "Acute Critical Warning Application System" achieved an Area Under the Curve (AUC) of 0.91 in predicting the critical condition and admission to the Intensive Care Unit (ICU) for hospitalized patients, validated through clinical field tests in collaboration with Landseed International Hospital.


Moreover, their collaboration with Taipei Veterans General Hospital on "Prediction of Heart Failure Patient Readmission and Mortality Risk," which incorporates X-ray, electrocardiogram, and electronic medical record information, demonstrated an average accuracy exceeding 0.90 and 0.92, respectively, in predicting the risks of readmission and mortality.

These innovative technologies provide strong support for early disease prediction and risk assessment.

MagTIES: Achieving Millisecond Precision in Neural Control with Innovative Magnetic-Electric Stimulation

The "MagTIES" developed by the Institute of Biomedical Engineering has also attracted attention. They utilize hexagonal disc-shaped magnetic-electric nanomaterials to achieve remote wireless neural control, boasting millisecond precision-a feat claimed to be unique globally-and possessing advantages such as low power consumption and low development costs.

Doctoral student Zhao-Jun Cheng from the laboratory mentioned that traditional magneto-electric stimulation employs bilayered materials, Under the influence of an alternating magnetic field, the inner magnetic core undergoes magnetostriction, stimulating the piezoelectric outer shell to generate an electric field for neural stimulation.

However, the fabrication of such materials is complex, costly, and offers time precision greater than one second. In contrast, "MagTIES" boasts exceptionally high precision in time and space, with lower development costs.

These research achievements once again highlight the significance of NYCU in biomedical engineering while also bringing boundless hope for future developments in medical technology.