WisdomEye Launches Shiye: The World's First Multimodal Healthcare Foundation Model, Accelerating Toward IPO

“In the AI era, all products deserve to be upgraded with large language models.”

“All future applications will be developed based on large models, and every industry should have its own large model.”

These two statements were made by Zhang Yong and Li Yanhong, respectively. The former was delivered at the summit where Alibaba’s “Tongyi Qianwen” large language model was officially unveiled, while the latter is excerpted from Li Yanhong’s speech at the 2023 Zhongguancun Forum.

Since the release of ChatGPT, a wave of large language models has swept across the globe. In March, Baidu launched “Ernie Bot”; in April, SenseTime introduced “SenseChat,” and Alibaba unveiled “Qwen”...

Nearly all major internet companies have joined the “battle,” with large language models becoming a “strategic battleground.”

However, for those outside the internet or technology sectors, there seems to be a common misconception that ChatGPT ignited the large language model (LLM) race. In reality, ChatGPT merely brought LLMs from behind the scenes into the spotlight, making them widely recognized by the general public.Long before the release of ChatGPT, an “arms race” in large language models had already erupted across various industries, such as entertainment and healthcare.

However,Regrettably, it is uncommon for companies to achieve both technological implementation and successful commercialization, and it is even rarer for them to successfully list on the secondary market.Therefore, companies that successfully achieve both technological implementation and commercialization have naturally become the focus of industry attention and serve as the key for external stakeholders seeking to gain a deeper understanding of this sector.

Athena Eyes Technology Co., Ltd. (hereinafter referred to as “Athena Eyes”) appears to be one such key.On May 20 this year, Athena Eyes officially launched its multimodal medical large language model, “Bianshi,” which empowers multiple healthcare scenarios, including pharmaceutical services, chronic disease management, and assisted diagnosis. Prior to this, Athena Eyes had achieved profitability for three consecutive years and had initiated work related to an initial public offering (IPO).

Much like the philosophical conundrum of “which came first, the chicken or the egg,” artificial intelligence companies face a similar dilemma: Is it more advantageous to pursue a development path that starts with technology and then seeks out application scenarios, or is it better to begin by identifying the needs within specific application scenarios and then refine the technology and products accordingly?

Athena Eyes has, from the very beginning,chose the latter,First, delve deeply into scenario-specific requirements, then develop tailored AI technologies and algorithmic models.etc.In this way, it not only paves the way for the practical implementation and commercialization of the technology in advance, but also better meets the needs of AI technology application in specific scenarios.

However, it is important to note that this does not mean Athena Eyes had no technological accumulation before seeking application scenarios. On the contrary,Before formally entering the healthcare sector, Athena Eyes Technology Co., Ltd. had already accumulated extensive experience in fields such as human resources and social security, medical insurance, and health administration. The company has built substantial expertise in core technologies—including computer vision, knowledge graphs, natural language processing, and privacy-preserving computation—maintaining a position at the international forefront.

For example, Athena Eyes’ facial recognition technology secured second place in Google’s MegaFace Challenge, a million-distractor algorithm competition, while its finger vein recognition technology claimed the world championship at the International Conference on Biometrics (ICB-CFVR) Global Algorithm Challenge for three consecutive years.

Building on its accumulated technological expertise, Athena Eyes has garnered the momentum to venture into the healthcare industry. Prior to developing specific artificial intelligence technologies and models, Athena Eyes conducted an in-depth analysis of the application requirements for AI within the healthcare sector, using these insights to refine its Bianstone Large Model.

Specifically, for China’s healthcare industry at the current stage, which problem demands the most urgent solution? Among all responses, the uneven distribution of medical resources and the pressing need to improve primary care capabilities undoubtedly rank first.

Indeed, in China, this issue has become an undisputed fact. The government has actively taken measures to promote the decentralization of high-quality medical resources and enhance primary healthcare capabilities. Policies such as tiered diagnosis and treatment, the establishment of medical alliances and medical consortia, vigorous promotion of hospital information technology infrastructure, and the implementation of telemedicine are all aimed at a common goal: addressing the uneven distribution of medical resources.

In this context, artificial intelligence technology is indispensable; it not only shoulders the critical task of enabling data interoperability but also helps primary care physicians enhance their clinical diagnostic capabilities through its auxiliary diagnostic functions.

Therefore,Primary healthcare has become the primary application scenario for Athena Eyes’ Bianstone Large Model. Medical record quality control, assisted diagnosis, and rational and standardized medication... The Bianstone Large Model will comprehensively empower primary care physicians from multiple aspects, including patient management and improving the quality of diagnosis, treatment, and medication.

It is important to emphasize that the original intention behind developing the Athena Eyes Bianstone Large Model-assisted diagnostic technology was not to replace physicians, but to serve as an intelligent tool for them. This technology aims to help physicians make more targeted diagnoses by combining their own past clinical experience with AI-assisted diagnostic recommendations, thereby enhancing the diagnostic capabilities of primary care physicians and facilitating the decentralization of high-quality medical services.

andWhile primary healthcare capabilities urgently need improvement, large public tertiary hospitals—the hubs of high-quality medical resources—are also grappling with issues such as patient overload, scarcity of physician resources, and excessively saturated workloads. The auxiliary diagnosis and rational drug prescription features offered by the Athena Eyes Bianstone Large Model can undoubtedly enhance physicians’ work efficiency to a certain extent and alleviate their excessive work burdens.

Moreover, personal growth is of paramount importance for physicians at all levels, from primary care providers to specialists in large public tertiary hospitals. To support this, the Athena Eyes Bianstone Large Model offers technical assistance for individual learning, such as medical knowledge retrieval, helping doctors continuously refine their medical expertise.

Besides hospitalsBeyond this scenario, Athena Eyes also targets three key application scenarios: physical medical and pharmaceutical institutions, chronic disease management, and health insurance fund supervision.

It is not difficult to understand the rationale behind deploying physical medical institutions and supervising the basic medical insurance fund, as they correspond respectively to pharmaceuticals and payment—two indispensable components of healthcare. As for chronic disease management and even broader health management, these represent a response to the emerging demand in recent years for full-lifecycle patient management.

So, specifically in the three scenarios of pharmaceuticals, chronic disease management, and payment, how can the Athena Eyes Bianstone Large Model provide empowerment?

For brick-and-mortar healthcare institutions, the demand for artificial intelligence (AI) technologies is more diversified. On the business front end, they require AI to provide intelligent support such as assisted prescription writing and auxiliary prescription review, as well as technical support for medical insurance settlement, drug delivery, and medication adherence tracking. On the business back end, these physical pharmaceutical institutions may also need digital management systems to provide managerial support in areas such as pharmaceutical supply chain management and pharmacy operations.

In this regard,Built on the Bian Stone large language model, Athena Eyes connects healthcare service demanders, providers, payers, and pharmaceutical suppliers, offering physical medical institutions services such as AI-assisted prescribing, AI-assisted diagnosis, intelligent medication management, and pharmaceutical care.

Moreover, physical medical institutions serve as key implementation settings for the concept of full-lifecycle health management for patients, which may also give rise to demands such as user retention, formulation of personalized health management plans, and collection and monitoring of user health data.

andLeveraging the Bianstone Large Model, Athena Eyes has developed a cloud-based chronic disease patient service management system that integrates in-hospital clinical data with out-of-hospital wearable device monitoring data. By employing AI algorithms and an evidence-based medical knowledge graph for intelligent data analysis, the system generates personalized chronic disease (health) management plans for each patient, thereby effectively supplementing post-diagnosis disease management.

However,It is worth noting that the deployment scenarios for this chronic disease management system, developed by Athena Eyes based on the Bianstone large model, are not limited to physical medical and pharmaceutical institutions but will also expand into home settings to meet emerging demands for family-oriented and personalized health management and medical services.

As for the layout of medical insurance supervision, it is not unrelated to the current significant pressure on China's medical insurance fund expenditures and the rapid growth of medical costs.

According to relevant statistics from VBInsight, from 2012 to 2020, the compound annual growth rate (CAGR) of expenditures from China’s urban and rural resident basic medical insurance fund was 36.56%, while the revenue growth rate during the same period was only 34%. In other words, the expenditure side of China’s basic medical insurance fund has been under sustained pressure.

In this context, China has continuously introduced corresponding medical insurance policies and begun to severely crack down on fraudulent activities involving health insurance fraud. Among these,By leveraging artificial intelligence and big data technologies, health insurance risk control models can be established to enable automated and intelligent analysis of the authenticity and compliance of medication purchases and diagnostic and treatment behaviors, thereby standardizing clinical practice and prescribing patterns, and achieving the goal of combating fraud and abuse in health insurance.

This is also the principle behind how the Athena Eyes Bianstone Large Model empowers the supervision of medical insurance funds and improves the efficiency of their utilization.

However, as previously stated, Athena Eyes distinguishes itself from other AI enterprises by first deeply exploring the requirements of application scenarios before developing specific artificial intelligence technologies and algorithmic models. In this context, what technical features does the Athena Eyes Bianstone Large Model possess compared to general-purpose industry large models, enabling it to fully realize Athena Eyes’ ambitious vision in the healthcare sector?

In the interview, Qiu Jianhua also provided a detailed explanation to VCBeat. First, a significant distinction between the healthcare industry and other sectors, such as entertainment and culture, lies in its strong emphasis on evidence-based practice. This means that medical decision-making requires a comprehensive consideration of clinical evidence, physicians’ experience, and patients’ individual circumstances and preferences, given that the accuracy of medical diagnoses has a direct impact on patients’ lives and health. Therefore, AI-assisted diagnostic systems must be grounded in authoritative medical evidence and clinical data, and must demonstrate high diagnostic accuracy.

In this regard,Athena Eyes’ Bianstone Large Model adopts a technical approach that integrates knowledge graphs with large language models. By incorporating knowledge graphs for knowledge enhancement during training and leveraging RLHF technology, the model achieves nearly a 10% improvement in the accuracy of medical question-answering compared to using large language model technology alone, all within a 50-billion-parameter framework.

Meanwhile, as medical data encompasses various unstructured forms, including text, images, videos, and audio, i.e.,Medical and healthcare data are multimodal. Therefore, to achieve the aforementioned functions such as intelligent consultation, assisted image interpretation, and health monitoring, the Athena Eyes Bianstone Large Model adopts multimodal visual processing technology. It supports the input of multimodal data in the medical field and enables the structured “output” of medical text for diverse tasks.

Not only that,Data security is of paramount importance in the healthcare sector, and the efficacy of artificial intelligence applications in this field depends not only on the scale of training data but also on its quality.In other words, for artificial intelligence (AI) technology to be applied in the healthcare sector in a compliant manner and achieve superior outcomes, AI enterprises must fully unlock the value of high-quality data while adhering to data security standards.

In this regard,Athena Eyes’ Bianstone Large Language Model addresses privacy-preserving distributed knowledge transfer and federated graph reasoning, enabling the aggregation of knowledge from distributed data silos. By employing federated large model training based on distributed data, it fully leverages the value of high-quality medical data.

Leveraging its three core technological advantages, Athena Eyes confidently projects that by the end of 2023, 5,000 village doctors, 20,000 pharmacies, and 1,000 blood purification centers will become partners in the Bianstone large ecosystem. Meanwhile, powered by the Bianstone Large Model, Athena Eyes will continue to empower primary healthcare institutions—including hospitals, pharmacies, and clinics—helping them enhance the quality and efficiency of medical services, provide patients with full-lifecycle health management, alleviate the burden of chronic diseases in key areas, and support the healthcare system in reducing costs and improving efficiency.

For hospitals, pharmaceutical institutions, and patients, the Bianstone Large Model by Athena Eyes offers numerous benefits. So, what does the Bianstone Large Model mean for Athena Eyes itself?

This signifies the initial success of shifting focus from the traditional human resources, social security, civil affairs, and health sectors to the broader healthcare industry, and also demonstrates Athena Eyes' determination to enter the healthcare market.

The “fruitful achievements” previously made in the artificial intelligence tracks of the human resources and social security, healthcare security, civil affairs, health, and financial sectors have naturally bolstered this determination with added confidence.

This achievement is,Athena Eyes has achieved profitability for three consecutive years, and its revenue in 2022 increased by 139.4% compared to that in 2020.

This is relatively rare among artificial intelligence (AI) enterprises. The reason is that, at the current stage, one of the major challenges facing many AI companies, including leading firms, remains escaping losses. Furthermore, a report released by An Hui, Secretary-General of the Artificial Intelligence Industry Innovation Alliance, indicates that more than 90% of companies in China’s AI product chain are still operating at a loss.

However, Athena Eyes not only achieved profitability for three consecutive years but also ranked among the top seven in China’s AI computer vision application market share. (Source: IDC, “Market Share of AI Computer Vision Applications in China (2020)”)

The achievement of consecutive profitability and a relatively high market share is not unrelated to the aforementioned strategy of identifying application scenarios before developing technical models; however, another primary factor is Athena Eyes’ unique market strategy.

VCBeat has learned that Athena Eyes drives its business with AI computing and adopts a “T-shaped strategy” for market expansion. Vertically, Athena Eyes leverages its autonomous, controllable, secure, and reliable AI technologies to deepen its engagement in the digital health industry. Horizontally, it is committed to expanding into diverse application scenarios, empowering governments, hospitals, citizens, and the entire industry to harness digital capabilities and advance the Healthy China Strategy.

Specifically, in the government sector, Athena Eyes has deployed products such as intelligent scenario monitoring, an AI middle platform for medical insurance, and big data anti-fraud solutions. For medical and pharmaceutical institutions, it has established service operation platforms covering chronic disease management, prescription circulation, primary healthcare, and end-to-end smart medical services. For the general public, Athena Eyes has built a highly competitive nationwide “Internet+” health and elderly care cloud platform.

In other words, by focusing on the upstream of the digital health industry supply chain and exploring diversification in the downstream market, Athena Eyes can leverage the synergy between upstream focus and downstream diversification, thereby creating a virtuous cycle of interaction.

As Athena Eyes Technology Co., Ltd. overcame numerous challenges, continuously enhancing its R&D capabilities, technological application implementation, and commercialization prowess, it launched an offensive into the secondary market, aiming to make a prominent debut.

“The War of a Thousand Models” has erupted across primary and secondary markets, as well as in the healthcare, entertainment, and other industries.