Healthcare systems across Asia have become more digital, but the movement of medical data remains uneven. Hospitals, clinics, radiology centers, insurers, and referral partners may already use electronic medical records, imaging systems, billing platforms, and digital workflow tools. Many of these systems, however, were designed around internal operations, which makes collaboration across institutions and borders harder than it should be.
The gap becomes more visible in cross-border care. Patient referrals can involve different languages, clinical documentation practices, medical standards, consent requirements, and data security expectations. A patient seeking specialist care in another market may still need records translated, summarized, reformatted, and manually reviewed before a receiving hospital can assess the case. For healthcare providers, these steps can create delays, duplicate work, and added risk in an already sensitive workflow.
Xiaoyan (Fiona) Zhou, Co-Founder and CEO of I.W.G, brings a useful perspective to this discussion because her work sits across several parts of the healthcare AI value chain. With nearly 10 years of experience in medical AI and digital healthcare, spanning business development, product strategy, hospital partnerships, regulatory collaboration, and international market expansion, she has seen how healthcare innovation depends on more than the availability of digital tools. Her earlier experience in AI medical imaging also shaped her view that adoption depends on whether technology can fit into the daily workflows of doctors, hospitals, and healthcare partners.
In this TNGlobal Q&A, Zhou discusses why interoperability remains difficult across Asia, how AI can support referral coordination and medical data exchange, and why infrastructure is becoming more important for healthcare AI adoption. She also outlines the guardrails needed around privacy, consent, auditability, data security, and human oversight as healthcare institutions explore more connected and AI-enabled medical data workflows.
Healthcare institutions across Asia continue to operate across fragmented systems, formats, and workflows. From your perspective, what are the biggest interoperability gaps that still slow down patient referrals and medical data exchange today?
The biggest interoperability gap is not simply the absence of digital systems. In many cases, hospitals already have electronic medical records, PACS, billing systems, or referral tools. The real gap is that many of these systems were designed for use within a single institution, not for collaboration across hospitals, clinics, insurers, and cross-border care networks.
Even when common standards such as FHIR exist, implementation in the real world is still highly fragmented. Different hospitals may use different vendors, different versions, different data structures, or different workflows. As a result, the information may be digital, but it is not always portable, searchable, or usable by another provider.
This is why patient referrals and medical data exchange are still often slowed down by manual processes, including fax, email, phone calls, and even physical media such as CDs for medical images. For clinicians, this creates duplicated work and delays. For patients, it means they may need to carry their own records between institutions.
So from my perspective, the largest gap is the lack of a practical interoperability layer that can translate between legacy systems, standardize data, and connect the actual referral workflow end to end.
Why has healthcare interoperability remained difficult to solve, even as hospitals, clinics, insurers, and healthtech providers have become more digitized?
Healthcare interoperability has remained difficult because digitization does not automatically create interoperability. Many hospitals, clinics, insurers, and healthtech providers have digitized their own operations, but they have done so using different systems, different data structures, different workflows, and different risk assumptions.
Healthcare is also a conservative industry when it comes to information sharing. Every organization wants access to more complete patient information, but at the same time, they are cautious about sharing their own data because of privacy, security, liability, and compliance risks. If something goes wrong, the cost is not only technical or financial — it can directly affect patient safety and institutional trust.
For example, we have seen a hospital that needed to use four different connection methods to communicate with around 100 partner clinics. Roughly every 20 to 30 clinics required a different integration approach. This shows that the challenge is not just whether a hospital is digital, but whether the digital systems can actually work together at scale.
So the problem is not only technology. It is a combination of legacy architecture, inconsistent standards, unclear incentives, and the lack of a trusted interoperability layer that can make data exchange safe, practical, and operationally useful.
Cross-border healthcare involves differences in language, clinical documentation, medical standards, and local regulations. How do these factors complicate the exchange of medical information across Asian markets?
Cross-border healthcare makes interoperability much more complex because the issue is not only whether data can be exchanged, but whether the information can be understood and trusted in a different clinical, linguistic, and regulatory context.
For example, imagine a cancer patient traveling from Indonesia to Japan for treatment. Traditionally, the patient’s medical records first need to be translated into Japanese by a medical translator. Then a medical tourism coordinator or clinical support team may need to summarize the case, identify suitable hospitals and specialists, and communicate with the receiving institution. When the Japanese doctor reviews the case, they also need to understand whether the diagnosis, test results, staging, treatment history, and clinical guidelines are interpreted in the same way as in the patient’s home country.
Every step takes time, and every manual handoff creates room for human error. This is where AI-enabled infrastructure can be very powerful. AI can translate, structure, and summarize medical documents, support hospital and doctor matching, and highlight differences in clinical criteria or guideline assumptions. Human professionals should still supervise and approve the output, but AI can take over much of the repetitive work and make cross-border medical data exchange faster, safer, and more scalable.
Many AI healthcare companies focus on diagnostics, clinical decision support, or patient-facing applications. Why is the infrastructure layer becoming increasingly important for the next phase of healthcare AI adoption?
Before starting this company, I spent five years working in an AI medical imaging diagnosis company. One of the most important lessons I learned is that the best AI is not always the one with the highest accuracy on paper. The best AI is the one that doctors can actually use naturally within their daily workflow.
For example, if an AI diagnosis or auto-reporting tool can save a doctor five minutes, but the doctor has to spend another five minutes exporting images, uploading data, switching systems, or copying results back into the medical record, then the value is lost. In healthcare, workflow friction can be the difference between adoption and rejection.
That is why the infrastructure layer is becoming so important. AI models need access to the right data, in the right format, at the right time, and they need to return results into the systems that doctors already use.
When we designed our platform, we assumed from the beginning that we would not build every AI model ourselves. Instead, we wanted to create an interoperable infrastructure where other AI partners can connect easily, deliver value through our workflow, and reach real clinical users without rebuilding the integration layer each time.
In markets with uneven digital maturity, how can healthcare organizations improve interoperability without requiring large-scale system replacement or costly custom integrations?
The short answer is: contact us.
The longer answer is that AI is changing the logic of system integration. Traditionally, interoperability has depended heavily on one-to-one API integrations. Two systems need to define the input, output, data format, and workflow in advance. But when the partner, use case, or local environment changes, the integration often needs to be redesigned or customized again.
AI allows us to think about this differently. When one person explains something to another person, and that person explains it again to a third person, humans do not need an API. We listen, understand the context, and restate the information in a way the next person can understand.
That is essentially what an AI referral agent can do for healthcare workflows. It can understand information from one institution, structure it, translate it, summarize it, and present it in the format another institution can use. As long as both sides have a basic digital format, organizations do not always need to replace their entire system or build expensive custom integrations.
This does not mean standards and APIs are no longer important. They are still necessary. But AI can act as a flexible interoperability layer on top of existing systems, helping markets with uneven digital maturity connect faster and more practically.
Patient data exchange raises concerns around privacy, security, consent, auditability, and regulatory compliance. What guardrails should healthcare organizations and healthtech providers put in place when adopting AI-enabled interoperability tools?
There is a common misconception in healthcare about what “security” really means.
I once had a meeting with a hospital where they told me paper documents felt safer because they could not be hacked. I asked them a simple question: if I quietly took one file from the desk, would you know which file I took? Would you know that I was the person who took it? In some cases, would you even know that the file was missing?
For me, real security starts with two fundamentals: permission control and access records. Healthcare organizations need to know who is allowed to access which patient data, under what conditions, and they need a complete audit trail of what happened.
For AI-enabled interoperability tools, the guardrails should be very clear. Data should be stored in secure cloud infrastructure, preferably within the relevant local jurisdiction when required. Access should be controlled at both the user level and the institutional level. Patient consent, data usage scope, and cross-border transfer rules should be clearly managed. Every action — who accessed which case, at what time, from which IP address, and what operation they performed — should be visible to authorized administrators.
At I.W.G, this is how we think about trust. We use leading cloud infrastructure such as AWS and Azure, design our systems around local data hosting where needed, maintain strict permission structures, and follow security and compliance frameworks including ISO 27001. AI can improve healthcare data exchange, but only when transparency, auditability, and human accountability are built into the system from the beginning.
Where are you seeing the strongest demand for cross-border medical data exchange in Asia today, such as medical tourism, specialist referrals, teleradiology, health checkups, insurance, or other use cases?
We are seeing strong demand in areas such as medical tourism, specialist referrals, advanced cancer treatment, health checkups, and second-opinion services. But behind these use cases, I think there is a deeper trend: medical resource mismatch.
Within every country, there is already a gap between large cities and rural areas in terms of access to specialists, advanced equipment, and high-quality care. If we expand that perspective across borders, the same logic applies at the regional level. Some Asian countries have fast-growing demand for advanced treatment, especially in areas such as oncology, while other markets, such as Japan, have highly developed medical capabilities and institutions that are looking for new ways to connect with international patients and partners.
Japan is one of the most advanced healthcare systems in Asia, with strong clinical expertise, high-quality medical infrastructure, and a long history of trust in medical services. At the same time, Japan is facing demographic challenges such as aging and population decline. This creates both pressure and opportunity for medical institutions to think beyond their traditional domestic patient base.
For us, cross-border medical data exchange is not only a technology problem. It is also a bridge-building problem. We help medical institutions, referral partners, and healthcare service providers exchange information more safely and efficiently, so that patients who need high-quality care can be matched with the right hospitals, doctors, and services across borders.
Looking ahead, how do you see AI changing healthcare interoperability and cross-border patient coordination in Asia over the next five years?
Over the next five years, I believe AI will change healthcare interoperability from a technical data exchange problem into an intelligent coordination problem.
Our vision is that when a doctor faces a difficult or rare case, they should be able to use our system to understand the latest relevant treatment options, compare similar cases, and even connect with doctors who have managed comparable patients before. The goal is not for a doctor to simply ask AI, “How should I treat this disease?” and receive a cold, generic answer. The real value is for AI to help organize knowledge, identify the right expertise, and create a trusted channel for collaboration between medical professionals.
For patients, AI-enabled interoperability can make the healthcare journey much clearer. Patients should be able to understand their condition, know where suitable treatment options are available, and be matched with the right hospitals and specialists across borders. Their medical records, images, summaries, and referral documents should move securely and smoothly, without unnecessary manual work or delays.
In Asia, where languages, systems, regulations, and medical resources are highly diverse, AI can become the coordination layer that connects fragmented healthcare networks. If we do this responsibly, with human supervision, security, and auditability built in, AI can help create a more accessible and collaborative healthcare system across the region.