2026/11/20

SG Biomedical (6623) announced on the 20th that it has partnered with the Hsinchu Branch of National Taiwan University Hospital (NTUH) and APMIC (Accelerate Private Machine Intelligence Company) to launch a development project for a medical-specific Large Language Model (Medical LLM). Utilizing the Hsinchu Branch of NTUH as the collaboration site, the initiative focuses on Traditional Chinese medical semantic understanding, clinical specialty applications, and healthcare data security, aiming to advance the development of localized medical AI technology in Taiwan.
SG Biomedical is set to list on the Emerging Stock Board (ESB) on November 23, with a reference price of NT$80 per share. The company has transitioned from semiconductor technology into the fields of innovative medical devices and AI-driven healthcare. Its current operations focus on two core areas: "Medical Cold Plasma" and "Medical Intelligence." In the short term, revenue is primarily driven by plasma medical equipment, consumables, and clinical services; in the medium to long term, the company aims to generate recurring revenue through subscriptions to medical AI models, licensing, and system integration services.
SG Biomedical stated that this collaboration will leverage the respective strengths of the three parties—spanning clinical medicine, artificial intelligence technology, and smart healthcare applications. Focusing on the specific needs of particular medical specialties, the partners aim to develop AI models tailored to Taiwan's unique healthcare environment. These models will undergo technical validation within actual clinical settings at medical institutions to explore the feasibility of future smart healthcare applications.
As global AI technology advances rapidly, large language models (LLMs) are emerging as a pivotal tool for the digital transformation of the healthcare industry. However, existing international LLMs are primarily trained on English-language data and global medical datasets. Consequently, adapting these models to Taiwan's specific context—including the use of Traditional Chinese in medical settings, the National Health Insurance (NHI) system, local healthcare regulations, and established clinical workflows—presents significant challenges regarding localization and regulatory compliance.
SG Biomedical notes that the medical field is highly specialized, with clinical needs and operational workflows varying across different medical specialties. Consequently, for medical AI models to be truly viable, they must not only possess specialized medical knowledge but also align with the actual operational environments of healthcare institutions.
This three-party collaboration focuses on developing a large language model (LLM) specifically designed for the medical sector. By integrating clinical practice, medical regulations, and hospital management workflows, the partners aim to build specialized models capable of understanding medical terminology and context in Traditional Chinese. This initiative seeks to overcome the limitations inherent in applying general-purpose AI models within Taiwan's healthcare environment.
Regarding the technical architecture, the development team plans to utilize an on-premises deployment model combined with autonomous data management by the healthcare institutions. This approach enables these institutions to conduct model development and application testing within their own existing IT management frameworks.
Given that medical data involves patient privacy and highly sensitive information, balancing the requirements for AI model training with data security and regulatory compliance has become a critical challenge in the adoption of medical AI. This collaboration integrates data security and privacy protection into the technical development roadmap, aiming to explore AI application models tailored to the specific needs of healthcare institutions.
Under the collaboration plan, the three parties will proceed with specialized model training, validation of medical domain knowledge, and testing of relevant applications. Through close cooperation between clinical settings and technical development teams, the initiative will evaluate the suitability of the models within the context of Traditional Chinese medical terminology and specific medical specialties, thereby continuously enhancing the professionalism and reliability of the medical AI systems.
SG Biomedical stated that the development of medical AI technology must align with clinical needs and real-world application environments. This partnership with the Hsinchu Branch of National Taiwan University Hospital and Accelerate Private Machine Intelligence Company (APMIC) will help bridge medical expertise with AI R&D resources, facilitating the exploration of potential applications for specialty-specific medical AI models.
Looking ahead, SG Biomedical remains committed to advancing smart healthcare technologies. By collaborating with healthcare institutions and AI technology providers, the company aims to accumulate experience in the development and validation of specialty-specific medical models—establishing a solid foundation for the future expansion of smart healthcare applications.
※ You are welcome to share the article link by "reposting" or "sharing"; please do not copy and paste the article content without authorization.