Fangxin Medical Files IPO Prospectus: Pioneering AI-Driven Digital Pathology Solutions Boosting Diagnostic Efficiency by 30%

In an era of rapid technological advancement, AI is reshaping industries at an unprecedented pace.As Wang Jianguo, founder of Fangxin Medical, stated in an interview with VCBeat: “AI will inevitably transform all industries and empower them as well.”Today, medical AI has long since moved beyond the laboratory to become a “super assistant” for physicians in the real world, reshaping diagnostic and treatment efficiency in hospitals.

Pathology departments represent one of the most mature scenarios for AI implementation. Traditional pathological diagnosis is a highly experience-dependent and labor-intensive “craft,” requiring pathologists to meticulously scrutinize biopsy slides with billions of pixels under a microscope to detect subtle clues. However, human vision is prone to fatigue, and expertise has its limits. In particular, training a qualified pathologist is akin to running a marathon, often taking ten years or more.

The penetration of cutting-edge technologies such as artificial intelligence and big data is driving a revolutionary transformation in pathology departments, shifting from “glass slides” to “digital images” and from “manual interpretation” to “AI-assisted analysis,” thereby redefining the operational models and development potential of the pathology industry.

In this context,Guangzhou Fangxin Medical Technology Co., Ltd. (hereinafter referred to as “Fangxin Medical”), established in 2017, has remained dedicated to providing comprehensive solutions for the development of digital and intelligent pathology departments. Fangxin Medical is not only the first enterprise in China’s pathology sector to propose the concept of “whole-process quality control and information management in pathology,” but also the first innovator to launch a “pan-disease AI-assisted diagnostic system,” thereby facilitating the transformation of the pathology industry from traditional diagnostic models toward greater efficiency, intelligence, and precision.

Pathological diagnosis serves as the core basis for clinical decision-making and is regarded as the “gold standard” for diagnosing the vast majority of diseases, particularly tumors. Its results directly impact every stage of patient care, including surgical planning, treatment strategies, prognosis assessment, and evaluation of therapeutic efficacy. As one of the most dynamically innovative fields globally, what exactly can AI bring to the department of pathology?

As a “veteran” with many years of experience in the pathology industry, Wang Jianguo believes that “the current major pain points facing the pathology sector are the significant shortage of pathologists and the uneven distribution of resources. Therefore, before Guangzhou Fangxin Medical Technology Co., Ltd. launched its AI R&D project for pathology, we discussed this issue with numerous pathology experts and further clarified the core value of AI in pathology departments—as a ‘powerful tool’ to enhance diagnostic efficiency.”

Generally, it takes a pathologist approximately 2–5 minutes to complete a pathology report for a simple case, and about 10–30 minutes for a moderately complex case. With the introduction of AI-powered pathology solutions, accurate diagnostic recommendations are generated and preliminary diagnostic reports are drafted by AI before physicians begin their diagnostic work. Consequently, physicians only need to review the diagnostic results and make minor adjustments to the report before issuance, significantly enhancing work efficiency.

In addition to accelerating report issuance,AI can also serve as a personal knowledge assistant for pathologists, helping them perform diagnostic tasks more efficiently and accurately.When pathologists encounter challenging images while examining pathological specimens, they can simply take a screenshot and send it to Fangxin’s multimodal intelligent Q&A assistant to promptly receive AI-based image analysis results and diagnostic recommendations. This significantly shortens the diagnostic turnaround time and helps prevent missed diagnoses due to limitations in clinical experience.

Meanwhile, AI can also provide an intelligent knowledge base that supports sustainable iterative upgrades.Wang Jianguo stated, “The variety of diseases currently known to humanity is extremely vast, with each disease having its unique morphology and manifestations. Therefore, it is unrealistic to expect pathologists to clearly memorize and accurately identify every single disease. In contrast, AI-powered intelligent pathology knowledge bases can continuously accumulate, iterate, and update, enabling physicians to automatically extract information from the database and efficiently search historical cases, thereby significantly improving the efficiency of pathologists in diagnosing complex and rare diseases, as well as in teaching, research, and learning.”

Based on the three core value propositions outlined above, Fangxin Medical’s independently developed AI-assisted pathological diagnosis system significantly reduces repetitive tasks for pathologists, markedly enhances work efficiency, and effectively addresses the industry-wide challenge of pathologist shortage. Currently, this product has been deployed in more than ten Grade 3A hospitals and has received accreditation from third-party laboratories.

Since 2024, the “Hundred-Model War” in the healthcare industry has intensified. Among these efforts, large pathology models have been hailed as the “crown jewel” of large language models due to their significant technical challenges and clinical value, prompting numerous companies and universities to launch a wide array of AI-powered products.

Wang Jianguo stated, “Currently, many large pathology models on the market are primarily designed for research settings, adopting a product design approach that assumes unlimited resources. They focus heavily on functional implementation while overlooking clinical application scenarios and resource constraints. In contrast, since initiating its project in 2022, Guangzhou Fangxin Medical Technology Co., Ltd. has remained committed to being guided by clinical needs and driving the continuous iterative R&D of pathology AI based on the requirements of limited resources.”

Following in-depth research, the team identified two core needs for pathologists that require urgent attention: the diagnosis of challenging cases and the automation of repetitive tasks.

First, the emergence of difficult-to-diagnose cases mainly stems from two factors. First, the cases themselves are extremely rare, lacking abundant prior diagnostic and treatment experience for reference; second, the boundaries for disease differentiation are blurred. In addition, various individualized factors such as the patient’s unique clinical manifestations, family medical history, and growth environment further increase the complexity of diagnosis. For this reason, when facing difficult-to-diagnose cases, the clinical intuition and comprehensive judgment of pathology experts have become the “gold standard” for accurate diagnosis.

Based on this, the pan-disease AI-assisted pathological diagnosis system developed by Fangxin Medical is designed to handle repetitive tasks in pathologists’ workflows and provide auxiliary diagnoses for most common diseases. By enhancing work efficiency, the system allows physicians to simply verify the accuracy of AI-generated results, thereby freeing up more time for diagnosing complex cases and acquiring new knowledge.

Dr. Chen Jintao, Head of AI R&D at Fangxin Medical, stated, “No matter how advanced the technology becomes, AI will never replace physicians, but it will profoundly transform healthcare delivery models. Therefore, by deeply integrating with the company’s whole-process pathology information management system, Fangxin’s Pathology AI can generate pixel-level heatmaps of suspicious lesion areas without disrupting physicians’ normal diagnostic workflows. This assists physicians in rapidly localizing lesions and verifying diagnostic accuracy, truly embodying the principle of ‘providing support without causing disruption, and collaborating without overstepping boundaries.’”

Throughout this process, humans and AI maintain a symbiotic and co-intelligent relationship. “AI employs deep learning on vast amounts of pathological images and diagnostic logic to construct a cognitive framework covering the characteristics of various diseases, providing physicians with real-time diagnostic clues and quantitative analytical support. Physicians, in turn, leverage their clinical experience to correct the AI’s reasoning pathways and optimize algorithmic models through feedback annotations, thereby forming a closed loop of ‘AI screening and diagnosis, physician confirmation and feedback, and AI secondary learning and verification,’” said Chen Jintao.

The integration of large and small models is another major highlight of Fangxin Pathology AI.

As a universal foundation for multimodal pathological knowledge, large language models (LLMs) exhibit strong generalization capabilities. However, when addressing highly specialized downstream tasks in medicine, their responses often lack precision and detail. Therefore, LLMs primarily provide broad conceptual knowledge of pathology and offer reference suggestions for complex cases.

In contrast, small models serve as specialized, precision executors in edge computing, focusing on local features at the task execution layer, such as fine-grained disease diagnosis, immunohistochemistry scoring, and disease grading. Through lightweight design, Fangxin Medical has reduced the computational load of small models by approximately 90%, enabling real-time processing of whole-slide images on standard GPUs and automatic localization of lesion areas via weakly supervised learning.

Finally, in response to the varying diagnostic needs of pathologists, Fangxin Medical has divided its pathology AI into discriminative models for qualitative analysis and generative models for creating pathology reports.Discriminative AI enables high-precision interpretation of pathological images across various organs, while generative AI facilitates natural language interaction to assist physicians in generating diagnostic opinions and recommending testing protocols.

By leveraging the synergy between large and small models, as well as the deep integration of discriminative and generative AI, Fangxin Medical’s independently developed pathology AI not only reconstructs the entire workflow of pathological diagnosis to enhance diagnostic efficiency, but also promotes the structured standardization and personalized content of pathology reports. This drives the field of pathology toward predictive and personalized medicine, providing more comprehensive support for physicians.

Leveraging these advantages, Fangxin Pathology AI achieves a diagnostic accuracy of over 90% across pan-organ applications. Furthermore, while maintaining a diagnostic sensitivity of 99%, it attains a specificity of approximately 92%.While ensuring no missed diagnoses, minimize misdiagnoses as much as possible, significantly enhancing the overall efficiency of the pathology department.

Research data from Global Growth Insights shows that the global digital pathology market reached $820.14 million in 2024. From 2024 to 2032, this market is projected to maintain robust growth at a compound annual growth rate (CAGR) of 19.31%. This rapid expansion has attracted numerous companies worldwide to enter the field. However, compared with their international counterparts, China’s AI pathology industry, despite its advantages of abundant proprietary databases and strong policy support, still lags behind in certain aspects.

For example, the digitalization of pathology departments is limited, with insufficient technological maturity; the massive data volume of digital slides entails high storage costs; and the inertia of traditional pathological diagnosis is difficult to overcome. In particular, China’s current training system for pathologists remains primarily based on microscopic slide examination, which has also constrained the rapid promotion and application of AI technologies in the field of pathology to some extent.

“Under the influence of numerous factors, how to break through the commercialization dilemma has become a major challenge facing the industry. This requires strengthened collaboration among all parties, including the government, medical institutions, universities, and enterprises, to jointly explore sustainable business models and drive the pathology industry’s leap into the AI era,” said Wang Jianguo.

In the face of a promising era of digital and intelligent pathology, Fangxin Medical has a clear strategic plan. Dr. Chen Jintao introduced:“Our AI products are planned along three dimensions: ‘breadth,’ ‘scope,’ and ‘depth.’ ‘Breadth’ refers to developing auxiliary diagnostic models that cover nearly all human organs. ‘Scope’ means ensuring data coverage across multiple centers and a wide variety of diseases. ‘Depth’ entails leveraging specialized small models tailored to hospital needs to perform more granular downstream tasks, such as histopathological grading, assessment of cancer invasion depth, and immunohistochemistry scoring. Ultimately, this approach aims to comprehensively enhance the coverage, disease recognition capability, and precision of pathological diagnosis.”

Wang Jianguo pointed out that to elevate the overall standard of the pathology industry, efforts should begin with enhancing the comprehensive efficiency of pathology departments, prioritizing the resolution of diagnostic efficiency issues. Leveraging efficiency improvements as a catalyst, the industry should advance its digitalization level, achieving critical qualitative breakthroughs through continuous quantitative accumulation. As the overall digitalization level of the industry rises, the inherent challenges of insufficient data and suboptimal quality for complex cases will gradually dissipate like melting icebergs.“We have reason to believe that prioritizing efficiency gains to encourage the adoption of pathological AI by hospitals and physicians, and then addressing the incremental innovation model for diagnosing complex cases once digital infrastructure reaches a new level, aligns with the objective laws of development. Meanwhile, pathological AI itself is continuously evolving; this step-by-step development path will inevitably accelerate its practical application.”

The 2025 Government Work Report emphasized the need to strengthen basic medical and health services and bolster the development of the pathology workforce. This policy direction has created a favorable environment for the robust growth of the AI-in-pathology industry. Against this backdrop, Wang Jianguo firmly believes:“We look forward to AI being comprehensively and deeply integrated into all aspects of pathology workflows within the next three to five years, through the concerted efforts of all stakeholders across the industry chain. This will significantly enhance diagnostic efficiency and effectively address the current challenges of prolonged training cycles, insufficient numbers, and uneven distribution of pathologists, thereby improving patients’ healthcare experiences and health outcomes.”