在恒指继续上冲，热门概念股普遍上涨的背景下，"医疗大模型第一股"讯飞医疗（02506）却要用业绩来证明其投资价值。 智通财经APP了解到，近日讯飞医疗（02506）发布2025年上半年财报，收入继续保持高增长水平，期间营业收入2.99亿元，同比增长30%，延续了2022- 2024年复合增速24.7%的高增长态势。这主要得益于该公司的研发驱动，研发成果化以及GBC三端应用场景协同发力，多元驱动收入不断创新高。 另外，该公司的盈利能力显著改善，一方面毛利率保持稳定，期间稳居51.5%高位，另一方面运营效率持续优化，销售与管理费用率下降7.5个百分点，净亏 损率显著收窄至22%，同比改善15.5个百分点。 实际上，讯飞医疗围绕"每个医生的AI诊疗助理，每个人的AI健康助手"，基于业内领先的人工智能技术和深度的行业认知，打造了用AI赋能政府（G）、医 院（B）、患者（C）的AI诊疗助理产品族和AI健康助手产品族，并形成了GBC三大场景驱动的多元模式。 2025年上半年，政府端业绩表现抢眼，收入1.4亿元，同比增长87%，占比收入47.4%。其中基层解决方案收入同比增长52%，区域解决方案增长178%。医院 及患者端 ...

# P a i g e 的筹码：稀 缺数 据与 合规落地 Paige成立于2018年，一直是全球病理AI的代表企业。它的价值不止在于团队，而在于同时握有数据、模型与合规三张稀缺的牌： ● 独家数据资源 ：依托纪念斯隆凯特琳癌症中心（MSK），Paige已积累近 700万张数字化病理切片及对应的临床数据 ，覆盖45个国家、多个癌种和不同人群。这 种数据规模和质量，在病理AI领域几乎无法复制。 北京和上海活动报名： 第三届全球手术机器人大会 2025医疗器械研发创新论坛 2025年8月22日，AI医疗公司 Tempus AI 宣布收购数字病理企业 Paige ，交易金额约 8125万美元（约合人民币5.8亿） ，主要以股票支付，并承担Paige与微软Azure 的剩余承诺。 从金额上看，这并不是一笔引人注目的"大单"。但在AI医疗产业的语境里，这却是一场足以改变棋局的交易。Paige带来了全球最大规模的病理数据资产之一，以及 首批获得FDA批准的病理AI应用；而Tempus，则正在构建肿瘤领域迄今为止规模最大的多模态大模型。两者结合，意味着 数据、临床和模型的拼图，正在快速拼合 。 Tempus创始人兼CEO E ...

Core Viewpoint - The company reported a significant increase in revenue and gross profit for the first half of 2025, driven by advancements in AI healthcare technology and supportive national policies [1]. Financial Performance - The company achieved an operating income of 299 million yuan, representing a year-on-year growth of 30.26% [1]. - The gross profit reached 154 million yuan, with a year-on-year increase of 26.96% [1]. Industry Context - The company is positioned to benefit from the "Digital China" and "Healthy China" strategies, with national policies promoting the integration of AI in healthcare [1]. - The company leverages its leading position in core medical model technology and extensive product coverage across the country to capitalize on the accelerating growth of the AI healthcare industry [1]. Future Strategy - The company plans to continue focusing on its development strategy, utilizing policy advantages to drive technological breakthroughs and deepen application scenarios in medical models [1]. - The company aims to provide higher quality medical services to residents and more efficient diagnostic tools for doctors [1]. Technological Advancements - Since the launch of the Xunfei Spark Medical Model in 2023, the company has expanded its technological advantages and achieved breakthroughs in multiple fields [1]. - In July 2025, the Xunfei Spark Medical Model underwent further upgrades, enhancing its capabilities and maintaining industry leadership in key medical task scenarios [1].

Core Viewpoint - Ping An Good Doctor (1833.HK) has demonstrated strong financial performance in the first half of 2025, indicating a sustainable business model and self-sustaining capabilities [1][2]. Financial Performance - The company achieved revenue of 2.502 billion RMB, a year-on-year increase of 19.5% [1] - Net profit attributable to shareholders reached 134 million RMB, up 136.8% year-on-year [1] - Gross margin improved to 33.56%, an increase of 1.37 percentage points [1] - Adjusted net profit under non-IFRS reached 165 million RMB, a growth of 83.6% year-on-year [1] Market Performance - The Hong Kong healthcare sector has shown strong performance, with the Hang Seng Healthcare Index rising 90% this year [3] - Ping An Good Doctor outperformed its peers, reflecting positive market sentiment towards its growth prospects [3] - Over the past 60 trading days, institutional investors have increased their holdings in Ping An Good Doctor, with a total increase of nearly 38 million shares [3][4] Business Growth - The company reported strong growth in both its F-end (financial client) and B-end (enterprise client) businesses, with revenues of 1.433 billion RMB (up 28.5%) and 527 million RMB (up 35.2%) respectively [6] - The number of paying users in the F-end reached approximately 20 million, a year-on-year increase of 34.6% [7] - The B-end health management business served over 3,500 paying enterprise clients, with a user growth of 39.2% [7] Market Opportunity - The aging population in China presents a significant market opportunity, with over 310 million people aged 60 and above by the end of 2024 [8] - The silver economy is expected to reach 30 trillion RMB by 2030, driving demand for healthcare services [8] - The company aims to leverage its integrated healthcare services through family doctors and elderly care managers to capture this market [8][9] AI Empowerment - The company is enhancing its service capabilities through AI, utilizing a closed-loop service model based on data, models, and scenarios [10] - AI-assisted consultation accuracy is approximately 98%, and the accuracy of complex disease treatment plans is nearly 80% [11] - The integration of AI is expected to significantly reduce service costs and improve operational efficiency [11]

Core Viewpoint - The public fund's strategy of "first doing drugs, then doing medicine" is enhancing the net value of pharmaceutical theme funds and creating switching opportunities in the market [1][4]. Group 1: Market Dynamics - In June and July, innovative drug theme funds began to double in value, accelerating the demand for portfolio adjustments among public funds, which in turn boosted interest in AI healthcare [1][3]. - By the end of June, many healthcare and technology theme funds reduced or completely sold off their positions in innovative drugs, reallocating those funds to AI healthcare stocks [1][3]. - As of August, AI healthcare stocks gained significant traction in the market, with several prominent fund managers predicting that AI healthcare would attract funds previously directed towards innovative drugs [1][10]. Group 2: Fund Performance - As of August 17, medical theme funds achieved a maximum return of nearly 150% in just eight months, primarily driven by the "drug" segment [3]. - Funds like Huatai-PineBridge and E Fund focused heavily on drug stocks, which contributed to their substantial returns, with some funds doubling their value within the year [3][4]. - The shift in focus from innovative drugs to AI healthcare is evident, with funds like Silver华 and 华夏 increasing their positions in AI healthcare stocks significantly by the end of June [6][8]. Group 3: Investment Strategy - Fund managers emphasize the importance of timing in the investment strategy, advocating for a sequential approach of "first doing drugs, then doing medicine" to maximize returns [4][10]. - The current phase of innovative drugs is characterized by a concentration of results and clear performance indicators, while AI healthcare is still in the early stages of product validation and commercialization [4][10]. - Fund managers are increasingly liquidating their positions in innovative drug stocks to switch to AI healthcare stocks, indicating a strategic pivot in their investment focus [8][10]. Group 4: Future Outlook - There is a growing expectation for AI healthcare to experience a rebound in the latter half of the year, with funds beginning to recognize its potential [9][10]. - Fund managers believe that AI healthcare could become a key investment theme by 2025, driven by collaborations between internet giants, pharmaceutical companies, and leading hospitals [10].

公募"先做药后做医"的布局节奏，正为医药主题基金带来净值提升与切换时机。 多位管理医疗主题产品的基金经理均强调，创新药与AI医疗是医药行业最大的两个新兴主题，创新药成熟在前，AI医疗 成熟在后，在医疗新技术、新产品上的布局必须考虑节奏，才能获得最大的收益空间，即"先做药后做医"。 鹏华基金权益投资二部副总经理、基金经理金笑非认为，2025年对创新药领域长期看好，是他目前重仓的方向，但也逐 步提高组合的进攻性，作为调仓思路，比如AI医疗。 今年6、7月份，创新药主题基金开始批量翻倍，使得公募基金的换仓需求加速释放，刺激AI医疗热度的提升。券商中国 记者注意到，6月末，多只医疗主题基金、科技主题基金减持甚至清仓创新药，并将减去的仓位替换为AI医疗股票。 8月份以来，伴随着一系列重磅基本面消息与机构资金博弈，AI医疗在股票市场热度明显加速提升，多位明星基金经理判 断，今年上半年AI医疗未能受益于创新药行情，考虑到AI医疗板块兼具医药属性与科技属性， AI医疗有望分流资金覆 盖，但也须关注主题投资弱化后的风险问题。 AI医疗赛道热度加速接棒 在主题行情轮动的节奏型行情中，尽可能在短时间博取最大收益，已是基金换仓需求的 ...

Core Insights - The Chinese government has approved a policy to further implement the "Artificial Intelligence +" initiative, emphasizing strong support for AI applications in various sectors, particularly healthcare [1] - Securities firms believe that AI applications in healthcare, such as AI pathology diagnosis, AI imaging, and AI pharmaceuticals, are maturing and may accelerate commercialization [1] - Companies in the healthcare sector are launching AI models, with notable examples including YiduCore by Yidu Technology, a clinical thinking model by Weining Health, and "Xihe No. 1" by Beijing Boya [1] Group 1: AI in Healthcare - AI is expected to become a strategic engine for a new wave of technological revolution and industrial transformation, with a focus on AI healthcare applications [1] - The "Xihe No. 1" model differentiates itself by using rigorous clinical data from hospital case files, replacing less reliable internet data [2] - The model has been developed over nearly 10 years, utilizing data from 18 medical institutions and 1 million real case studies, achieving a coverage rate of 98% and a precision rate of over 90% in medical knowledge [2] Group 2: Addressing Healthcare Challenges - The "Xihe No. 1" model aims to address high misdiagnosis rates (over 25%) and delays in treatment, which can increase mortality risk by 35% [3] - A strategic partnership has been established to enhance AI-driven health management in Ningxia, focusing on improving access to healthcare in remote areas [3] - The company aims to bridge the gap in medical resources, allowing patients to receive expert-level guidance through AI [3] Group 3: Future Directions - The company plans to expand its AI healthcare services, targeting over 100 million patient interactions within three years [4] - Future developments will include AI smart terminal hardware products such as smart wheelchairs and nutritional management devices [4] - The establishment of a global medical AI collaboration network is intended to enhance model accuracy and facilitate global sharing of medical wisdom [4]

2025年8月15日至8月17日，第31届中国国际医用仪器设备展览会暨技术交流会(China-Hospeq)在北京国 家会议中心举办。本届展会以"共迎健康挑战，共谋发展机遇"为主题，展会面积22000平方米，参展企 业超过300家，展会重点展示了医学影像、医学检验、放射治疗、AI医疗大模型、数智化解决方案、生 命支持、卫生应急等多领域中国制造创新成果。 这种数据选择的差异，直接决定了模型的可靠性。据介绍，"羲和一号"的训练数据均来自18家医疗机构 的临床一线，涵盖人口学特征、时序理化指标、影像组学等多模态信息，且每一条数据都经过伦理审 查、脱敏处理，标注过程由临床医生全程参与。 "互联网数据可能包含健康资讯、用户问答，但这些内容缺乏临床验证，甚至存在错误。"栗琳举例说， 用互联网数据训练的模型好比"隔着书本学看病"，而"羲和一号"是在临床一线"跟着医生查房问诊"。 数据"质地"的差距体现在具体性能上："羲和一号"的幻觉(错误输出)率极低，准确率可达90%以上；而 部分依赖互联网数据的模型，可能出现"普通咳嗽误判为严重肺病"等致命错误。"医疗领域差10%的准 确率，就是对患者健康的巨大风险。"栗琳强调，守住"临 ...

Group 1 - The 31st China International Medical Instrument and Equipment Exhibition and Technical Exchange Conference will be held in Beijing from August 15 to August 17, 2025, featuring a seminar on AI in healthcare [1] - The seminar aims to leverage AI technology to create a "Health Silk Road," breaking down barriers in medical resource allocation and enhancing the efficiency of healthcare services across regions [2] - Multiple initiatives were launched during the seminar, including the establishment of a global children's early diagnosis and treatment AI innovation ecosystem, contributing to the construction of the "Health Silk Road" [2] Group 2 - The "Xihe No. 1" medical model, developed by Peking University Third Hospital and other partners, has been a focal point, utilizing 1 million real clinical cases and featuring over 100 billion parameters [3] - The model's training data comes from 18 medical institutions, ensuring high accuracy and compliance, with a medical knowledge coverage rate of 98% and precision rate exceeding 90% [4] - The model aims to address high misdiagnosis rates in cardiovascular diseases and improve diagnostic efficiency in primary care settings [5] Group 3 - The company is building an AI + healthcare ecosystem that encompasses data, models, and clinical applications, focusing on rapid translation of research outcomes into practical applications [6] - The company is also exploring AI-enabled hardware products to meet healthcare service demands, indicating a broadening of its service offerings [6] - The company has formed strategic partnerships to enhance data quality and application in AI medical models, aiming to serve over 100 million patients within three years [8][9]

Group 1 - The 31st China International Medical Instrument and Equipment Exhibition and Technical Exchange Conference will be held in Beijing from August 15 to August 17, 2025, featuring a seminar on AI in healthcare [1] - The seminar aims to leverage AI technology to create a "Health Silk Road," breaking down barriers in medical resource allocation and enhancing the efficiency of healthcare services across regions [2] - Multiple initiatives were launched during the seminar, including the establishment of a global children's early diagnosis and treatment AI innovation ecosystem, contributing to the construction of the "Health Silk Road" [2] Group 2 - The "Xihe No. 1" medical model, developed by Peking University Third Hospital and other partners, has been highlighted, featuring over 100 billion parameters and a 98% coverage rate of medical knowledge [3] - The model's training data is sourced from 18 medical institutions, ensuring high accuracy and compliance, with a reported error rate of less than 10% [4] - The model addresses significant issues in cardiovascular disease diagnosis, where traditional methods have high misdiagnosis rates [5] Group 3 - The company is focused on building an AI + healthcare ecosystem that spans various fields, including omics research and drug development, to accelerate the translation of research outcomes [6] - The company is also exploring AI-enabled hardware products to meet healthcare service demands [6] - The company aims to serve over 100 million patients within three years and establish a global AI collaboration network for shared medical wisdom [9] Group 4 - The company, Tianjin TEDA Biomedical Engineering Co., Ltd., is committed to optimizing medical resource allocation through AI technology [8] - The company has formed strategic partnerships to enhance data quality and application in AI medical models [8] - The company plans to continue its exploration in the AI + healthcare sector, collaborating with various stakeholders to improve healthcare access and quality [9]