当美国巨头如 Google 、 OpenAI 和 Anthropic 竞相开发支撑其 AI 产品的大型语言模型时， Sakana AI 、 Mistral AI 、 DeepSeek 和 AI21 Labs 等初创公司正凭借为特定地区、行业或独特功能设计的专业模型开辟自己的细分市 场。 为实现这一目标，总部位于东京的 Sakana AI 已完成 200 亿日元（约合 1.35 亿美元）的 B 轮融资，公司联合创始 人兼 CEO David Ha 向 TechCrunch 透露，本轮投后估值达 26.5 亿美元，较投前 25 亿美元的估值有所提升。 Sakana AI 由前谷歌研究员 Llion Jones 、 Ren Ito 和 David Ha 于 2023 年联合创立，致力于开发性价比高的生成式 AI 模型，这些模型适用于小数据集，并为日语及日本文化进行了优化。 此轮 B 融资吸引了新老投资者共同参与，既包括日本金融巨头三菱 UFJ 金融集团（ MUFG ），也有 Khosla Ventures 、麦格理资本、 NEA 、 Lux Capital 和 In-Q-Tel （ IQT ）等全球风投机构。 Ha ...

Core Insights - Artificial Intelligence (AI) and Value-Based Care (VBC) are transformative forces in the healthcare sector, aiming to enhance precision medicine, control rising healthcare costs, and improve patient experiences and outcomes. However, both face challenges in implementation within existing healthcare systems [1] Group 1: AI as an Integrator - A key strategy for VBC is the formation of integrated healthcare systems that enhance value through seamless coordination of multidisciplinary opinions and community services. However, data diversity and fragmentation hinder this integration, negatively impacting healthcare quality. AI can act as a "real integrator" across data sources and systems, utilizing generative AI models to extract and interpret vast amounts of heterogeneous data [2] Group 2: Enhancing Alternative Payment Models - Changing payment methods is fundamental to enhancing value in healthcare. Various Alternative Payment Models (APMs) exist, but their adoption in developed countries like the U.S. is slow due to uncertainties regarding patient intervention needs and the transparency of necessary services. AI can address these issues by predicting optimal intervention timings and identifying cost drivers in patient care journeys, thereby promoting the use of APMs [4] Group 3: Patient-Centric Outcome Measurement - Traditional healthcare outcome measurements often do not align with what patients truly care about. AI can improve the efficiency of gathering patient feedback and standardize subjective feedback into actionable decision-support tools. This shift may lead to more frequent use of patient-centered outcome measures in health technology assessments, influencing the pricing of new drugs and devices based on the value they provide to patients [5][6] Group 4: Overcoming Fragmentation Challenges - To realize the potential of AI in healthcare, it is essential to overcome data fragmentation and ensure diverse datasets are available for training models that deliver real patient value. AI models must be validated using recognized standards across diverse populations, and optimizing payment for AI products in a value-based system is crucial [8] Group 5: Mutual Empowerment of AI and VBC - The diverse AI tools and applications in healthcare present a risk of fragmentation during implementation. VBC offers a strategic framework to systematically deploy AI capabilities, creating a high-value healthcare system. Thus, AI and VBC empower each other, working together to fulfill their revolutionary promises and usher in a much-needed new era in healthcare [9]

【环球网科技综合报道】6月14日消息，据路透社报道称，消息人士透露，谷歌本周已与 Scale AI 的多家竞争对手接洽，意图转移大部分业务。Scale AI 失 去了大量业务，而 Meta 的入股使其估值达到 290 亿美元，高于交易前的 140 亿美元。 消息人士称，Scale AI 2024 年收入达 8.7 亿美元，谷歌去年在 Scale AI 的服务上花费约 1.5 亿美元。除谷歌外，微软、埃隆・马斯克的 xAI 也计划退出 Scale AI。OpenAI 数月前决定退出，尽管其支出远低于谷歌。OpenAI 首席财务官表示，公司将继续把 Scale AI 作为众多数据供应商之一。 多家公司担忧与 Scale AI 合作可能使研究重点和路线图暴露给竞争对手 Meta。客户通常会共享专有数据及原型产品，鉴于 Meta 持有 49% 股份，竞争对手 担心其商业战略和技术蓝图可能被知晓。 谷歌一年多来一直在寻求数据服务提供商多元化，而 Meta 的举动促使谷歌加速退出 Scale AI。由于数据标签合同的结构，这一过程可能迅速发生，为 Scale AI 的竞争对手提供了机会。 然而，由于其核心业务依赖少数关键 ...