Asustek Computer (Asus) has been accelerating the integration of AI, hardware, software, and cloud technologies into healthcare as it builds a next-generation smart healthcare ecosystem. Kuo-Hao Huang, vice president of the Asus AI Research Center, said the biggest challenge hospitals face in adopting AI is not the AI models themselves, but whether their data, workflows, and system architectures are ready.
Asus is building around its next-generation healthcare information system, xHIS, combining AI, hybrid cloud, a modular microservices architecture, platform-based development tools, and standardized healthcare data models aligned with international standards to create an AI-ready digital core. The goal is to make healthcare data and workflows easier to integrate, standardize, and scale.
Rather than simply upgrading conventional hospital information systems (HIS), xHIS is designed to bring AI from standalone tools into clinical workflows and serve as the digital foundation for smart hospitals, community healthcare, and integration across healthcare institutions.
The following are excerpts from DIGITIMES' interview with Huang.
Q: What challenges do existing hospital HIS platforms face when adopting AI, and how does xHIS address them?
Huang: Traditional HIS platforms support patient registration, consultations, physician orders, examinations, billing, insurance claims, and medical record management. When hospitals introduce AI, however, they face fragmented data and difficulties embedding AI into existing workflows.
For example, a patient's medical history, medications, laboratory results, allergy records, imaging, and nursing data may be distributed across different systems and formats. AI cannot make accurate judgments if that information has not been organized into a consistent, comprehensible format that can be accessed in real time.
Another problem is the difficulty of integrating AI directly into clinical workflows. An AI-powered medication alert, for example, cannot consider only medications recently prescribed by a physician. It should also incorporate the patient's medical records, allergy history, and other medication information. If these systems cannot communicate with one another, AI is reduced to a standalone add-on.
Asus xHIS Digital Health Platform (DHP) does more than add AI capabilities to an existing HIS. It is designed around the needs of next-generation smart healthcare institutions, rebuilding the digital core to make it AI-ready.
The platform combines AI, hybrid cloud, a modular microservices architecture, and platform-based development tools while adopting standardized healthcare data models aligned with international standards. This makes healthcare data and workflows easier to integrate, standardize, and scale.
The architecture also accommodates hospitals' diverse clinical requirements, reducing the complexity of customization and system integration while improving the efficiency of smart healthcare deployment. This makes it easier for AI to move beyond individual applications and become part of actual clinical workflows.
Q: What challenges arose during xHIS deployment at New Taipei City Hospital, and what has been achieved since launch?
Huang: The deployment at New Taipei City Hospital represents an important milestone for Asus. A hospital information system cannot simply be shut down and replaced; migration has to be completed without interrupting healthcare services. One of the challenges was reorganizing years of hospital workflows into a standardized, modular, and platform-based architecture.
Deploying xHIS was not simply a matter of installing software. We worked with the hospital to restructure its core workflows while ensuring that the system remained stable, secure, and capable of supporting real-world healthcare operations.
xHIS is now being used across the hospital's core operations, supporting major workflows including outpatient, emergency, and inpatient services. This demonstrates that xHIS has effectively become the digital core underpinning both hospital operations and clinical workflows.
Because xHIS adopts a three-tier architecture, modular design, standardized data foundation, and open APIs, subsequent applications—including AI-assisted medical records, intelligent consultations, patient services, and care solutions—can be integrated in a more standardized manner without requiring extensive customized integration each time.
New Taipei City Hospital has therefore done more than complete an HIS project. It has established replicable smart healthcare infrastructure.
Going forward, we hope to extend this platform to community healthcare integration, chronic disease management, and cross-institution care, connecting these services with home healthcare and long-term care. Starting from core hospital workflows, xHIS can gradually expand into broader healthcare scenarios involving AI applications and cross-system integration.
Q: AI is evolving from generative AI toward agentic AI. How does this differ from previous AI applications in healthcare?
Huang: Generative AI has traditionally helped produce medical records, summaries, patient education materials, and administrative documents—essentially providing information or generating content. Agentic AI goes a step further by understanding context, breaking down tasks, connecting with tools, and helping carry out subsequent actions within its authorized scope.
This is particularly sensitive in healthcare. Once an AI agent begins affecting physician orders, medications, patient communications, care workflows, or clinical decisions, issues involving patient safety, medical liability, and regulation arise. Asus is therefore taking a relatively cautious approach to agentic AI.
In the first stage, we position agentic AI as a copilot rather than an autopilot for healthcare teams. AI can help organize information, flag risks, recommend next steps, and handle administrative tasks, but critical medical decisions and accountability remain with healthcare professionals.
The second stage focuses on low-risk, auditable, and reversible workflows, such as medical record summarization, speech-to-medical-record conversion, patient education materials, administrative documents, patient services, and anomaly detection. Human confirmation, access controls, and operation logs are retained throughout the process.
Once governance mechanisms mature, agentic AI can move into higher-value clinical workflows. For example, it could flag potential risks in physician orders based on laboratory results, allergy histories, and medication records. Even then, human-in-the-loop mechanisms must remain in place, supported by traceability, monitoring, and accountability.
Agentic AI cannot simply function as an external add-on. It must understand healthcare data, integrate with clinical workflows, recognize user permissions, and record every action. Through standardized data, modular workflows, open APIs, access controls, and audit logs, xHIS is designed to ensure that AI agents operate safely, under appropriate controls, and within a governable framework.
Asus is also incorporating AI agents into AI-native engineering for xHIS itself, allowing agents to participate in requirements gathering, system design, software development, testing and validation, troubleshooting, and operational monitoring. This extends AI across the entire software lifecycle. Generative AI produces content and insights; AI agents participate in workflows and actions.
Q: What other technologies or applications does Asus plan to introduce into smart healthcare?
Huang: Asus will not focus solely on individual AI applications. With xHIS DHP at the core, we aim to establish platform capabilities that can support the long-term evolution of smart hospitals.
First, xHIS will integrate scenarios spanning outpatient, emergency, inpatient, nursing, pharmacy, laboratory, health examination, and long-term care services to establish a common data foundation.
Second, we will advance more AI-native clinical workflows, including medical record summarization, speech-to-medical-record conversion, physician order assistance, nursing care plan recommendations, personalized patient education, chronic disease and community care management, patient safety reporting, and clinical risk alerts. These applications are intended to reduce administrative workloads for healthcare professionals while improving care efficiency and patient safety.
Third, we will develop an open healthcare application ecosystem. xHIS will continue evolving toward open APIs, Fast Healthcare Interoperability Resources (FHIR)-aligned architecture, SMART on FHIR, modular applications, and platform governance, enabling hospitals and external partners to develop and deploy services on a secure, compliant, and controlled foundation.
Fourth, Asus will strengthen hybrid cloud, sovereign cloud, cybersecurity, privacy, and AI governance. Innovation in smart healthcare must be accompanied by security, accountability, and effective governance.
Healthcare will increasingly extend beyond hospitals to clinics, communities, homes, and long-term care environments. Asus ultimately wants xHIS to become the digital backbone of Asus Healthcare, allowing smart healthcare to expand from in-hospital workflows into AI, community healthcare, and long-term care while creating a more continuous and integrated care ecosystem.
Article translated by Scarlett Yu and edited by Ysi Chen
