Taiwan's NHI Data Is Thorough. But Foreign Experts Still Can't Use It: How I Built a Trilingual Catastrophic Illness Lookup Tool from One ODT File

TL;DR: Taiwan’s NHI open data is thorough and well-structured, but a multilingual, interactive entry point has yet to appear. As regenerative medicine becomes a global priority, that missing interface layer is a gap worth filling. I started with what I needed for my own work: turning the Ministry of Health and Welfare’s 30 NHI catastrophic illness categories into a trilingual Chinese-English-Japanese tool page, and opening a Regenerative Medicine & Technology content series on paulkuo.tw.

Where things began and where they ended up are quite different.

In 2024, a Taiwanese medical technology company approached Sunpoint International hoping to bring Taiwan’s cell factory construction and operations services to Japan. The project drew on the accumulated factory-building expertise and ecosystem of Taiwan’s semiconductor supply chain: cleanroom standards, process validation, equipment supplier networks, and the discipline of construction teams. They had developed an integrated facility design suited to cell therapy, and the goal was to turn it into a turn-key system that could be transplanted to Japan for production.

After more than a year of exploration, the project evolved beyond simply exporting Taiwanese services to Japan. It gradually became a deeper form of Taiwan-Japan collaboration. By mid-2025, both sides were planning to establish a Japanese-affiliated subsidiary in Taiwan, with the goal of bringing a specific cell therapy approach for a hard-to-treat disease to market there. Sunpoint International’s role in this project is to coordinate the Taiwan-side work. Chairman Jerry Lee (李建樑) serves as the primary liaison; I joined as Chief Strategy Officer and partner.

Our medical consultant on the project is Taiwanese by birth but grew up in Japan. He works in Japanese and English. He does not read Chinese, so Taiwan’s NHI and pharmaceutical regulatory materials come with an extra language barrier for him.

We often prepare materials on his behalf: how Taiwan’s NHI “catastrophic illness” categories are classified, which version of ICD-10 applies in Taiwan, how long a certificate for a given diagnosis remains valid. These are the foundational facts that need to be established before any pharmaceutical application can proceed in Taiwan.

Then I opened the NHI’s official “Catastrophic Illness Categories” page.

The Chinese version is complete: all 30 categories, ICD-10-CM codes, certificate validity periods, and a full history of regulatory amendments. Six download links sit on the page, in PDF, ODT, and ODS formats, all with matching content.

There is an English counterpart page, titled Patients with Catastrophic Illnesses or Rare Diseases, but it covers mainly the conceptual framework: what catastrophic illness means, how to apply, and what benefits follow. For the Japanese medical specialists we work with, the conceptual framework is understandable but not granular enough. What they need is a lookup table: which code does this diagnosis map to, which major category does it fall under, how long is the certificate valid. That kind of searchable, cross-referenceable detail currently exists only in the Chinese files.

There is no Japanese counterpart page at all.

So every time I needed to answer a question from the Japanese side, I would download the Chinese ODT, scroll to the third page to find the relevant entry, translate that passage from Traditional Chinese into Japanese or English, and paste it back into LINE or email.

The data is technically findable. The problem is that once found, it still requires a human to translate and reformat it before it can actually be used in a cross-language collaboration.

The Data Is There. The Interface Is Not.

By any measure of completeness and open-format availability, the NHI’s data disclosure already ranks near the top in Asia.

The “Catastrophic Illness Categories” page has had a complete version since 2015 (Year 104 of the Republic of China calendar), updated with every regulatory revision. The current version was revised on September 16, 2024 (Year 113) and took effect January 1, 2025 (Year 114), aligned with the 2023 edition of ICD-10-CM/PCS. Six historical versions are kept in the same folder, from the pre-2015 edition through to the latest, usable directly for institutional research or year-over-year comparison.

Format-wise, there is a PDF for easy reading; ODT and ODS for editing. All 30 categories, with ICD-10-CM codes, certificate validity periods, and subcategory breakdowns, are presented in structured form.

In other words, the NHI has already built a solid foundation in terms of content and structure.

What is genuinely missing is the interface layer: multilingual, interactive, shareable.

If what you need is a page where you can type in a diagnosis term, immediately see the corresponding catastrophic illness number, validity period, and classification chapter, and then switch the display to English or Japanese to share with a collaborator, that page does not exist yet.

A Global Shift in How Public Data Gets Used

Stepping back, this is not a problem unique to any one country.

The way people retrieve information is changing in the AI era. Ten years ago we searched Google; five years ago we consulted Wikipedia. Now a growing share of people simply ask ChatGPT, Perplexity, or Claude. The next generation of users may never open an official website first. They will ask an AI directly, and that AI needs to draw answers from multilingual, structured, interactable open interfaces.

This creates the same pressure on public data everywhere: the shift from static downloadable files to machine-readable, multilingual, interactive interface layers. Japan’s Ministry of Health, Labour and Welfare; the U.S. CMS; the UK’s NHS; Singapore’s MOH; Health Canada, all are at different stages of the same challenge: making public data not merely public, but genuinely usable across languages and professional backgrounds. No country has finished this work.

Taiwan has real advantages here. Its single-payer NHI produces a level of data integration rare globally. ICD-10 codes are tightly linked to the catastrophic illness, special care, and reimbursement frameworks, with consistent internal logic. Policy also moves quickly. The Regenerative Medicine Act and the Regenerative Medicine Preparations Act were both promulgated in June 2024 and took effect January 1, 2026, a legislative and implementation timeline rare anywhere in the world.

The point is not who is ahead or behind. This global evolution of data interfaces is still ongoing, for everyone.

ICD-10 Is Not Something You Look Up on Wikipedia

My introduction to ICD-10 did not come from an encyclopedia. I studied life sciences as an undergraduate, and before that I trained as a medical laboratory technologist. This disease classification system ran on the hospital computers I worked with. The International Classification of Diseases, Tenth Revision has more than 20,000 codes. Each letter prefix designates a body system: A for infectious diseases, C for malignant neoplasms, I for the circulatory system, M for musculoskeletal and connective tissue disorders, Q for congenital malformations. That structure is part of how I think.

Taiwan’s NHI links ICD-10 directly to the catastrophic illness framework. The same diagnosis, placed under different chapters, can carry a different certificate validity period, a different scope of copayment exemption, and a different reimbursement pathway. That layer of local institutional design reflects thirty years of accumulated NHI logic.

The NHI makes all of this publicly available, PDF, ODT, ODS, with matching content across formats.

But no matter how complete the content, if a collaborator does not read Chinese, someone still has to extract the relevant information from those files and convert it into a language and format the other party can actually work with.

From One ODT File to a Usable Entry Point

Why start with the 30 catastrophic illness categories? Because this is the exact document I had already worked through for a real job requirement. Downloading the ODT, cross-referencing ICD-10, mentally organizing the 30 categories into a structured table: that process was itself a good opportunity to learn how Taiwan’s NHI system is designed.

But once I was done, all of that distilled understanding lived in my head and in a few email drafts. If it had value for my collaborators, it should have value for the next Taiwanese professional returning from Japan, the next BD evaluating investment from the United States, the next family member who just received a diagnosis and wants to know what benefits they are entitled to, the next regulatory consultant preparing a pharmaceutical submission.

The tool page is not meant to replace the NHI’s official data. It sits on top of what the government has already built and adds a multilingual, interactive entry point. Links to the original source files will be prominently placed on the tool page so users can always return to the Ministry of Health and Welfare website to verify the latest version.

This is only the first brick. More can be added.

Why Open This Series

Why start a Regenerative Medicine & Technology series at all?

Professionally: the Regenerative Medicine Act and the Regenerative Medicine Preparations Act were promulgated in June 2024 and took effect January 1, 2026. The rules of the game for regenerative medicine in Taiwan are being recalibrated, and the industry is entering a new starting phase. Search in English or Japanese, though, and you will find that Taiwan’s regulatory framework, industry developments, and key data in this space still lack a complete, accessible entry point. As healthcare, regenerative medicine, and longevity services become global priorities in the post-AI era, Taiwan has sufficient industrial and institutional foundations to participate meaningfully. The accessibility of the relevant information, however, remains underdeveloped.

Personally: I began working with exosome and cell therapy-related services in 2024, which led me to think about this field more often from three angles simultaneously, as user, family member, and industry participant. Personal experience is not clinical evidence, and I have no intention of writing testimonials. What is worth discussing is what kind of knowledge infrastructure Taiwan needs as this industry becomes institutionalized, data-driven, and international.

There is also a longer arc. When I first entered life sciences and medical laboratory work thirty years ago, Taiwan’s biotechnology sector was at the “about to take off, but not quite yet” stage. Looking back now, threads that once seemed unconnected, life sciences, medical technology, AI, cross-border collaboration, have converged around this topic. If the industry is entering a growth phase, it will need more entry points that readers from different backgrounds can understand, search, and use.

The Regenerative Medicine & Technology series on paulkuo.tw starts here. I will continue adding pieces that are useful in my own work and that I hope others will find useful too. If you are interested in building personal data infrastructure, the earlier piece on personal health data infrastructure applies the same logic at the individual scale. And if you have ever been stuck, as I was, by a document that is technically public but practically inaccessible across languages, this series may be worth following.

This is where the first article ends.

Glossary

• ODT (OpenDocument Text): An open-standard word processing format. Similar to .docx but under an open license; can be opened with LibreOffice, Google Docs, or Microsoft Word.
• ODS (OpenDocument Spreadsheet): An open-standard spreadsheet format. Similar to .xlsx but under an open license; can be opened with LibreOffice, Google Sheets, or Microsoft Excel.
• PDF (Portable Document Format): A cross-platform document format designed primarily for reading. Content is fixed and difficult to edit.
• ICD-10-CM/PCS (International Classification of Diseases, 10th Revision, Clinical Modification / Procedure Coding System): The disease diagnosis and procedure coding foundation currently used by Taiwan’s NHI. ICD-10 is the classification system published by the World Health Organization (WHO); ICD-10-CM is a clinical modification maintained primarily by the U.S. NCHS and CMS; ICD-10-PCS is used for inpatient procedure classification and is published by the U.S. CMS. Taiwan’s NHI has adopted and localized the 2023 edition of ICD-10-CM/PCS.
• Catastrophic Illness (重大傷病): The 30 disease categories designated in Schedule 1 of the Regulations Governing Exemption from Co-payment for NHI Beneficiaries, Article 2. Holders of a catastrophic illness certificate are exempt from copayments for related treatments.
• Regenerative Medicine Act (再生醫療法): The law governing the administration of cell therapy techniques by medical institutions. Promulgated June 2024; effective January 1, 2026. Governs the clinical delivery side.
• Regenerative Medicine Preparations Act (再生醫療製劑條例): The law governing the research, manufacture, and marketing of regenerative medicine products, including cell therapy drugs, by pharmaceutical companies. Promulgated June 2024; effective January 1, 2026. Governs the pharmaceutical manufacturing side. The two laws together are known as the “regenerative medicine dual legislation.”
• Turn-key system: A fully integrated, ready-to-operate system delivered as a complete package.
• GMP (Good Manufacturing Practice): The international quality standard governing the production of pharmaceutical and medical products.

Note: This article is a synthesis of public data and institutional observation. It does not constitute medical, legal, or NHI application advice. Eligibility for catastrophic illness status, certificate validity, and applicable scope should always be verified against the latest announcements from the National Health Insurance Administration, Ministry of Health and Welfare, physician diagnosis, and the determinations of competent authorities.