Okay, diving into the fascinating intersection of ancient wisdom and cutting-edge technology here in China! It’s wild to see AI popping up everywhere, from predicting traffic jams to composing music. But one area that’s particularly intriguing, and maybe a little mind-bending for folks back home, is its foray into Traditional Chinese Medicine, or TCM.
I run this blog to give you a slice of life and the big trends happening here in China, and this one feels like a big deal, maybe a game-changer, or maybe… well, let’s dig in. We’ve got whispers of IPOs, cautionary tales of sudden collapses, serious debates about technology, and even robots attempting to take your pulse. So, let’s explore this “AI + TCM” trend – what’s the real story? Is it the future of healthcare or just a lot of hype?
The Ancient Practice Meets the Algorithm Age
For many Americans, TCM might conjure images of acupuncture needles, herbal teas that smell… earthy, and perhaps complex theories about ‘qi’ and meridians. It’s a medical system with a history stretching back thousands of years, deeply rooted in philosophical concepts and extensive clinical experience, often passed down through generations from master to apprentice. Unlike Western medicine, which often focuses on specific diseases and biological mechanisms, TCM takes a holistic view, looking at the body as an integrated system and seeking to restore balance.
This reliance on deep experience, subtle observation (“望闻问切” – looking, listening, asking, and pulse-feeling), and highly individualized treatment plans (“千人千方” – a thousand people, a thousand prescriptions, meaning prescriptions are tailored) is both its strength and, in the modern world, its challenge.
And this is where AI enters the chat.
The tech world here in China, much like globally, is buzzing with the transformative power of Artificial Intelligence, particularly large language models (LLMs). The narrative is that AI can “re-do” every industry. Medicine, especially drug discovery and diagnostics in Western medicine, has seen significant AI investment. But applying AI to TCM? That seems like a whole other ball game, fraught with unique difficulties given its qualitative, experience-based nature. Yet, companies and institutions are diving in headfirst.
The “Why” Behind the AI Push in TCM
Why is AI seen as a potential savior for this ancient practice? The articles I’ve been reading point to some pretty fundamental pain points in the traditional TCM system here in China.
First off, quality control is a big one. Like many healthcare systems globally, China faces a significant shortage of highly skilled practitioners. Specifically for TCM, becoming a truly expert practitioner takes years, often decades, of dedicated practice and mentorship. The best doctors are in high demand, leading to long wait times and difficulty accessing top-tier care, particularly in rural or less developed areas. Reports indicate that senior TCM physicians make up a small percentage of the total, far below the demand.
Secondly, the dispersed nature of patients and the need for in-person consultation makes scaling challenging. While online consultations exist, the core diagnostic methods of “望闻闻切” traditionally require physical presence.
Thirdly, while rich in historical texts and case studies, TCM suffers from a certain lack of standardization and data structure when compared to modern Western medicine. Clinical records aren’t always uniform, descriptions of symptoms and syndromes can vary between different schools of thought (and there are many!), and translating the nuanced observations of a skilled practitioner into quantifiable data points is, shall we say, tricky. This “black box” aspect of TCM, particularly regarding the complex chemical interactions in herbal formulas, makes it hard to explain, replicate, and gain wider international acceptance using current scientific paradigms.
This is where AI is being pitched as a solution. By leveraging AI’s ability to process vast amounts of data, identify patterns, and potentially standardize diagnostic approaches, the hope is to:
- Scale Access: Augment the capabilities of less experienced practitioners, potentially increasing the effective supply of high-quality TCM services.
- Standardize Knowledge: Capture and disseminate the accumulated knowledge of master practitioners and historical texts in a more structured, accessible way.
- Modernize and Validate: Apply modern data science techniques to explore the underlying mechanisms of TCM therapies and standardize manufacturing of complex herbal products.
- Improve Efficiency: Automate tedious tasks like medical record writing, freeing up doctors for patient care.
Basically, the idea is to take the deep, but often hard-to-scale and standardize, knowledge base of TCM and “level it up” with the computational power and data-processing capabilities of AI.
What Does AI-Powered TCM Look Like on the Ground?
So, how is this playing out in practice? It ranges from somewhat whimsical consumer gadgets to serious clinical tools and massive R&D projects.
You might encounter what some are calling “Cyber TCM” robots in hospitals or public health centers. One article describes them vividly: “Its eyes are cameras, scanning your tongue coating and complexion like an old TCM doctor; it can also take your pulse, a sensor terminal completing the reading in less than 2 minutes.” These often produce a quick health report based on basic diagnostics and pattern matching – maybe more for fun or initial screening, but showing the public face of AI in TCM.
But the more serious efforts are focused on building sophisticated AI systems and large models designed to assist actual practitioners and potentially even drive new drug discovery.
Companies are developing AI-assisted diagnostic systems. AskZhen (问止中医科技), a company making waves by attempting an IPO in Hong Kong, centers its business around its proprietary “TCM Brain” (中医大脑). This system, they claim, is an AI auxiliary diagnostic system built on the “world’s largest clinical knowledge graph” for TCM. The idea is that a doctor inputs patient data (symptoms, diagnostic findings like tongue/pulse info), and the AI processes this against its massive database of historical cases, classical texts, and modern research to suggest potential diagnoses, syndromes, and treatment plans (including herbal prescriptions). AskZhen claims their system has passed clinical consistency studies at a top-tier hospital and expert review. They see it as a core tool enabling their doctors to provide “in-depth and comprehensive consultations” with less dependence on individual experience, effectively boosting the quality of care they can provide.
Beyond diagnosis, AI is being applied to TCM big models for research and development. Companies like Tasly Pharmaceutical (天士力), a major player in traditional Chinese medicine, have partnered with tech giants like Huawei Cloud to launch models like “数智本草” (roughly, “Digital Herbal Wisdom”). These models are “fed” gargantuan amounts of data – thousands of ancient texts, tens of thousands of traditional formulas, millions of research abstracts, natural product data, patent info, clinical guidelines (Huxiu 1 article details the scale for “数智本草”: over a thousand ancient texts, 90,000+ formulas, 40,000+ patented formulas, 40 million+ literature abstracts, millions of natural products and target pathways, etc., totaling 38 billion parameters). The goal? To “explain the rationale” (“说理”) behind traditional formulas and “find drugs” (“寻药”), accelerating the discovery of new natural compounds or combinations and enabling a “數智中药” (Digital TCM) R&D paradigm. This represents a significant shift, with traditional pharma companies reportedly boosting R&D spending considerably, investing hundreds of millions of RMB into digital upgrades of manufacturing lines and building these massive data systems (Huxiu 1 article gives an example of a 300M RMB intelligent extraction workshop).
Hospitals are also getting in on the action. The prestigious Guang’anmen Hospital (广安门医院), often referred to as the “Peking Union Medical College” of TCM, recently launched its “广医·岐智” (GuangYi QiZhi) big model. This model is being integrated throughout the hospital’s workflow – in outpatient clinics, inpatient wards, even surgery. It aims to assist doctors with everything from patient interviews and medical record writing (reportedly generating notes in seconds) to multi-modal analysis using diagnostic images (like tongue scans) alongside verbal input. The hospital leadership explicitly states the model is not meant to replace doctors but to “copy TCM experts, improve hospital efficiency, and free up doctors from琐事 (琐事 means trivial matters or chores), allowing them to focus on service quality.” Crucially, they also see it as a tool to accelerate knowledge transfer and training for younger doctors, effectively allowing AI to act as a “digital master,” helping to “flatten out” variations in skill level based on individual physician talent and experience.
Capital Flocks In: The IPO Attempt and the Market Buzz
This convergence of ancient practice and modern tech has certainly caught the eye of investors. The articles highlight increased capital flow into “AI + TCM” startups. We’ve seen early-stage funding rounds announced for companies working on AI models for health management and diagnostics using digital “望闻闻切” (36Kr AI+TCM article mentions examples like Guiyuan Hall and WuZheng Intelligent).
The most prominent recent example is AskZhen’s (问止中医科技) bid to list on the Hong Kong Stock Exchange. This is a significant move, positioning them as a potential “first AI TCM stock.” Founded in 2018, interestingly, by three individuals with Silicon Valley and investment backgrounds rather than deep roots in traditional TCM academia, AskZhen quickly gained traction, raising several rounds of funding from prominent VCs like Bluerun Ventures.
Their financials show a company in rapid growth mode: revenue jumped from about 62 million RMB in 2022 to 189 million RMB in 2023, and reached 173 million RMB in the first nine months of 2024 (TMTpost figures). This growth is primarily driven by their medical services, which accounted for almost 90% of their revenue in 2024 Q3, delivered both online and through their expanding network of physical clinics across China. Their “TCM Brain” subscription service for other practitioners and institutions is a much smaller revenue stream (under 5% in 2024 Q3, per TMTpost).
However, despite the revenue growth, AskZhen has also reported substantial net losses – around 154 million RMB in 2022, 194 million RMB in 2023, and 56 million RMB in the first nine months of 2024. That’s a cumulative reported loss of roughly 400 million RMB over this period (36Kr AI+TCM article highlights this). They do report positive adjusted net profit (5.15 million RMB, 43.03 million RMB, and 49.51 million RMB respectively, adjusting for non-cash items like share-based compensation – TMTpost figures), a common metric for tech companies showing underlying operational profitability before certain expenses.
AskZhen claims to be the largest AI-enabled TCM service provider in China by 2023 revenue, but their market share is still only 1.5% in a highly fragmented market with over 12,600 providers (TMTpost figures). This suggests both significant growth potential and intense competition.
The market size predictions further fuel the excitement. The overall TCM medical service market in mainland China was estimated at around 960 billion RMB in 2023, projected to grow to 1.7 trillion RMB by 2028 (12.3% CAGR). The AI-enabled segment, while much smaller (around 10.9 billion RMB in 2023), is projected to grow at a blistering 51.4% CAGR to 86.9 billion RMB by 2028 (TMTpost figures). These numbers paint a picture of a rapidly expanding niche within a large, growing market.
But Wait, There’s a Catch (or Several)
The narrative of rapid growth and AI-driven transformation sounds compelling, and the investment activity is real. But like any gold rush, there are significant risks and challenges that temper the hype. The articles I read were quite candid about these.
1. The Data Problem is Fundamental: This is perhaps the biggest hurdle. As mentioned, TCM data is inherently difficult to standardize. How do you quantify the feeling of a pulse (“slippery,” “wiry”) or the subtle variations in tongue color or coating? Different doctors describe the same phenomena differently. Different schools of thought approach diagnosis and treatment from fundamentally different theoretical frameworks (e.g., the “Fire Spirit” school vs. the “Warm Diseases” school). Feeding AI models messy, inconsistent, or highly subjective data limits their accuracy and reliability. While companies like AskZhen and hospitals like Guang’anmen are trying to create structured databases and “knowledge graphs,” it’s a monumental task. The “千人千方” principle also means a vast amount of variability that’s hard for pattern-matching algorithms to perfectly replicate.
2. AI’s Limitations vs. Human Expertise: Can AI truly replicate the diagnostic intuition of a skilled TCM doctor who combines book knowledge with years of hands-on experience, patient interaction, and environmental observation? Many experts are skeptical. They argue that AI can analyze data points but struggles with the holistic, nuanced “thinking chain” of a human doctor, especially in complex or rare cases. There’s also the issue of “AI hallucination,” where models generate plausible-sounding but incorrect information – potentially dangerous in a medical context. While models are getting better, and techniques like augmented retrieval from curated knowledge bases help, the risk isn’t zero. The Huxiu 2 article mentioning a GPT-4 test on TCM exam questions showed a far lower accuracy rate (43.9%) compared to human practitioners (70-78%), specifically attributing errors to misunderstanding TCM concepts and relying on a Western medical framework.
3. Business Model & Execution Risk: Is the AI truly the core value proposition driving revenue, or is it primarily a marketing angle to attract investment and patients? Looking at AskZhen’s financials, the AI subscription revenue is tiny compared to their medical service revenue, which is generated by… human doctors. While the “TCM Brain” might make those doctors more efficient or consistent, it’s still the doctor-patient interaction that’s paying the bills. Also, AskZhen’s relatively low R&D spending compared to marketing spending (36Kr AI+TCM article points this out) raises questions about how much cutting-edge AI development is really happening versus how much is spent acquiring customers.
4. Legal, Ethical, and Patient Acceptance Issues: If an AI suggests a prescription that leads to a negative outcome, who is responsible? The doctor who approved it? The company that developed the AI? The legal framework is unclear. This concern is already leading to regulatory caution; the 36Kr AI+TCM article mentions Hunan Province explicitly banning the use of AI to automatically generate prescriptions due to drug safety concerns. Furthermore, patient trust is paramount in healthcare. While people might enjoy the “Cyber TCM” novelty, are they willing to trust an AI for serious medical conditions, especially given the personalized nature of TCM? An article mentioning a patient complaint about worsening cancer after using AskZhen’s services serves as a stark reminder of the potential stakes and controversies involved.
5. The Cautionary Tale: Eagle Eye’s Collapse: Not every AI + TCM story is about IPO ambitions. The sudden and dramatic dissolution of Eagle Eye (鹰之眼), a company claiming over 600 employees and widespread adoption of its AI-powered infrared diagnostic devices, provides a sobering counterpoint. Reports detail employees being notified of the company’s dissolution via a late-night message, followed by unpaid wages, unfulfilled severance promises, and mass labor arbitration. Employees questioned the actual technical sophistication of their products, citing instances of identical diagnostic reports from repeated scans. The company was described as having internal chaos, redundant departments, internal competition instead of collaboration, and top-down decisions driven by the founder’s ideas rather than expert input (like buying expensive, unused equipment). Their struggles to sell products – reportedly offering free devices but seeing low patient self-pay uptake for the associated tests – suggest a significant mismatch between their technology and market viability. Eagle Eye’s failure is a concrete example of how hype, poor execution, and financial unsustainability can quickly lead to collapse in this space, reinforcing the “bubble” concern.
Transformation or Bubble: The Million-Dollar Question
So, is “AI + TCM” the future or just a speculative bubble? Based on the reports, it seems it’s likely a bit of both, or rather, a complex, dynamic space where potential, hype, and significant challenges coexist.
The market opportunity is real, driven by genuine needs in China’s healthcare system and government support for TCM modernization. The technological potential of AI to process complex data and assist in pattern recognition is also undeniable. Companies and institutions are making real investments and developing sophisticated systems.
However, the fundamental nature of TCM – its reliance on subjective observation, individualized treatment, and a theoretical framework different from the empirical science underlying most AI training data – presents unique and difficult technical hurdles. Simply “feeding” AI a lot of data isn’t enough; the data itself needs to be standardized and structured in a way that respects the core principles of TCM, a task that is still very much in progress.
The Eagle Eye case shows the risk of startups failing due to poor management, questionable technology, and inability to find a viable business model. AskZhen’s IPO attempt highlights the capital markets’ excitement but also exposes underlying questions about its profitability, R&D depth, and reliance on human doctors, suggesting the AI might currently be more of an enhancement tool and marketing differentiator than a complete replacement or primary revenue driver.
Ultimately, many experts agree that AI is unlikely to replace experienced TCM doctors anytime soon, if ever. The human element of diagnosis, empathetic interaction, and nuanced decision-making in the face of uncertainty remains crucial. As one article quotes an expert saying, “What will replace TCM doctors is not AI, but TCM doctors who use AI” (“替代中医的不是AI，而是会用AI的中医”).
The future likely involves AI acting as a powerful assistant or tool (“數智化基建” – digital infrastructure) for TCM practitioners, helping them process information, improve efficiency, standardize certain aspects of care, and potentially uncover new insights from historical data. It could help train younger doctors faster, making high-quality care more accessible. It could revolutionize drug discovery and manufacturing consistency.
But navigating the challenges of data quality, ethical considerations, and building sustainable business models in this complex, culturally-specific domain means the path forward for AI + TCM will be anything but straightforward. It’s a high-stakes experiment blending ancient wisdom with modern technology, and whether it leads to a true transformation of TCM or a burst bubble remains a fascinating story to watch unfold from here in China.
Hope this gives you a clearer picture of this intriguing development! It’s definitely one of those things where the reality is much more complex than the headlines might suggest. Let me know what you think in the comments!
评论