来自伦敦大学学院（UCL）计算机科学系的研究团队提出了一种用于白细胞形态分析的联邦学习框架，使各机构能够在不交换训练数据的情况下进行协同 训练。利用来自多个临床站点的血液涂片，该联邦模型在保证完全数据隐私的同时，学习到稳健且域不变的特征表示。与集中式训练相比，联邦训练在跨 站点性能和对未知机构的泛化能力上表现出色。 血液形态学检查是临床诊断血液疾病的重要环节，通过观察外周血涂片（PBS）或骨髓穿刺（BMA）中的细胞形态，医生可以判断白血病、贫血、感染 及遗传性血液疾病的类型。然而，这一过程不仅劳动强度大，而且高度依赖经验丰富的专业人员。尤其在低收入和中等收入国家（LMICs），技能专家稀 缺，使得快速、可靠且可扩展的血液学诊断成为急需解决的问题。 近年来，人工智能和深度学习的发展为血液形态分析提供了新的解决方案。AI 模型能够自动识别不同类型的白细胞，并辅助医生进行快速诊断。研究表 明，深度学习在自动化血液学诊断中具备显著潜力，但现实应用中仍面临重要挑战——模型训练对数据的依赖性极强，而临床数据通常分布在不同医院， 且存在染色方法差异、成像设备差异以及少数罕见细胞类型的问题。这种数据异质性会导致模型在新机构或 ...

Core Insights - The report by PitchBook emphasizes that AI is transforming the biopharma industry significantly, with efficiency improvements being a key characteristic [1][18]. Group 1: Market Activity - In 2025, the AI drug development sector is projected to complete 101 transactions, totaling $2.7 billion [2][19]. - AI-enabled pharmaceutical tools and services account for 62.5% of the total transactions, while AI-native biotech companies have increased their capital attraction to 70.1% [5][22]. - The median transaction size for AI-native biotech is expected to reach 1.9 times the industry median in 2024 and 1.6 times in 2025 [10][27]. Group 2: Valuation and Investment - AI-native biotech companies achieved a median valuation of $78 million during financing in 2024, nearly double that of the broader biopharma industry [7][24]. - The share of AI drug development in total capital has been rising, reaching 9% in 2024 and 8.4% in 2025, despite maintaining a transaction volume share of around 4.1% over the past five years [10][27]. Group 3: Clinical Success Rates - AI-native biotech companies report clinical success rates of 80-90% in early asset development, significantly higher than the industry average of 40-65% [11][28]. - In Phase II trials, the success rate aligns with historical averages at 40%, indicating potential for improved efficiency in drug development without increasing costs [11][28]. Group 4: Business Models - A subset of AI drug startups operates as service providers rather than developing drugs themselves, generating valuable learning data and providing access to AI platforms [12][29]. - These service-oriented companies can achieve revenue in a shorter timeframe, although they face challenges from large multinational companies building their own platforms [12][29]. Group 5: Investment Trends - Over the past four quarters, venture capital activity in AI-native biotech has increased significantly, with transaction counts rising by approximately 34% [14][31]. - The average transaction size for these platforms is $6.8 million, demonstrating higher capital efficiency compared to traditional drug development companies [35].

Core Insights - Owkin, an AI company focused on biological complexities, has announced a collaboration with NVIDIA to enhance its biological large reasoning model, OwkinZero [1] Group 1: Collaboration Details - The partnership will utilize NVIDIA's advanced AI computing ecosystem, which includes the Nemotron family of open models and the NVIDIA NeMo framework [1] - The collaboration aims to improve the performance, scalability, guardrails, and robustness of Owkin's biological reasoning capabilities [1] Group 2: Objectives - The primary goal of this collaboration is to accelerate advancements in biological reasoning, which is critical for the company's mission [1]

位于波士顿的生物技术公司Insilico Medicine似乎是首家将新一代人工智能（称为Transformer模 型）应用于药物研发领域的公司。早在2019年，该公司的研究人员就思考能否利用这些模型，根据 生物和化学数据研发新药。他们的首要目标是特发性肺纤维化，一种肺部疾病。 他们首先利用与该疾病相关的数据集训练人工智能，并找到了一种很有前景的靶蛋白。随后，第二 个人工智能推荐了一些能够与该蛋白结合并改变其行为，但毒性和稳定性又不太高的分子。之后， 由化学家接手，合成并测试了筛选出的分子。他们将最终成果命名为rentosertib，该药物近期已成 An AI revolution in drugmaking is under way 它将彻底改变药品的生产方式，以及整个行业本身。 阿特里克·施瓦布并非普通的药物研究员，他的工作场所也并非普通的制药实验室。这里既没有实验 台，也没有冒泡的液体，更没有白大褂。施瓦布博士总是身着一身黑衣。但这对于他位于国王十字 区的工作场所来说，倒也十分贴切。国王十字区曾是铁路货场和工业区，如今经过改造，已成为伦 敦最令人艳羡的时尚街区之一。 施瓦布博士就职于葛兰素史克（GSK） ...

Core Insights - Outcomes4Me, a medical technology company based in Massachusetts, raised $21 million in funding in May 2025, led by Salica Investments, with participation from Labcorp Venture Fund and Forecast Labs [1] - The total funding raised by Outcomes4Me to date amounts to $38 million (approximately 270 million yuan), which will be used to further develop its AI-driven cancer care platform aimed at providing personalized and intelligent solutions for patients [1] Funding History - Outcomes4Me has undergone several funding rounds: - Seed Round in November 2020: $4.7 million from Asset Management Ventures [2] - Series A in April 2021: $12 million from Northpond Ventures [2] - Series A+ in May 2025: $21 million from Salica Investments [2] Market Demand and Challenges - The global cancer incidence is rising, with 19.976 million new cases and 9.744 million deaths reported in 2022, leading to increased demand for cancer care [3] - Current challenges in cancer care include low patient treatment adherence, limited access to clinical trial information, and fragmented data across multiple platforms, which affects treatment continuity [3] AI Cancer Care Platform Features - Outcomes4Me's AI cancer care platform includes three core functionalities: - Personalized treatment recommendations using AI and machine learning to analyze patient medical records [4] - Clinical trial matching to improve patient participation rates by integrating data from over 700 medical institutions [4] - Symptom tracking through an app that allows patients to record symptoms and share data with healthcare teams, ensuring data security in compliance with U.S. medical privacy regulations [4] Competitive Landscape - Similar AI-driven cancer care products include PathAI's AI Sight imaging management system and Owkin's MSIntuit CRC digital pathology solution, which aim to enhance patient understanding and treatment optimization [5] Industry Trends in China - China's cancer incidence is also on the rise, with 4.8247 million new cases reported in 2022, accounting for 24.1% of global cases [6] - The Chinese government is promoting AI applications in healthcare, with guidelines released in November 2024 outlining 13 major categories and 84 subcategories for AI technology in the medical field [6] - Several AI-driven cancer care platforms have emerged in China, such as CSCO AI and Ant Group's AI health manager, indicating rapid development in this sector [6] Future Outlook - Hospitals are increasingly adopting "AI + nursing" technologies, with initiatives like the integrated "symptom assessment-AI matching-VR intervention" terminal launched by Sun Yat-sen University Cancer Center [7] - The demand for continuous information and support for cancer patients post-discharge is critical, suggesting a growing need for AI-integrated solutions in cancer care [7]

Financial Data and Key Metrics Changes - Revenue in the fourth quarter was $0.7 million, reflecting progress in partner programs [37] - Research and development expenses increased to $18.4 million for the three months ending December 31, 2024, compared to $12.3 million for the prior year period, driven by advancements in internal programs [37] - Selling, general, and administrative expenses decreased to $8.8 million for the three months ending December 31, 2024, from $9.3 million in the prior year period [38] - The company ended the year with $112.4 million in cash, cash equivalents, and short-term investments, down from $127.1 million as of September 30, 2024 [38] Business Line Data and Key Metrics Changes - The ABS-201 program targeting androgenic alopecia is positioned as a potential flagship asset, with preclinical data suggesting high affinity and potency [19][21] - ABS-101, an anti-TL1A antibody, showed promising preclinical data indicating reduced internalization of TL1A complexes and prolonged target engagement [24][25] - ABS-301 and ABS-501 programs also demonstrated significant preclinical results, with ABS-301 showing potent anti-tumor response and ABS-501 confirming novel epitope interactions [26][27] Market Data and Key Metrics Changes - The market for androgenic alopecia is substantial, affecting approximately 80 million people in the US, with a significant unmet need for innovative treatments [20] - The company is exploring partnerships in the immuno-oncology and animal health markets, indicating a broad interest in various therapeutic areas [30][32] Company Strategy and Development Direction - The company aims to develop ABS-201 internally through later-stage clinical development to retain maximum value [23] - A strategic collaboration with AMD, including a $20 million investment, is expected to enhance the company's AI drug creation capabilities [15][39] - The focus is shifting towards a balanced portfolio of high-quality therapeutic programs, emphasizing internal programs and selective partnerships [33][34] Management's Comments on Operating Environment and Future Outlook - Management expressed confidence in the ability to execute across the portfolio, with expectations to become a clinical-stage biotech company as ABS-101 enters the clinic [44] - The company anticipates signing new partnerships, including with large pharma, to enhance its drug creation capabilities [36][44] - The existing cash reserves are projected to fund operations into the first half of 2027, supporting ongoing and future clinical studies [42][102] Other Important Information - The company has added four new partners to its ecosystem, enhancing its collaborative capabilities [8][29] - The focus on Generative AI for antibody design aims to create differentiated candidates and unlock novel biology for patient benefit [17] Q&A Session Summary Question: Plans for additional preclinical studies with ABS-101 and ABS-201 - Management confirmed that additional preclinical studies for ABS-101 are planned, with data to be disclosed at scientific conferences [51][52] Question: Updates on partner program developments - Management emphasized a focus on large pharma partnerships and potential out-licensing of ABS-101, aiming for significant upfront payments to extend cash runway [56][58] Question: Design and endpoints for the planned Phase I study of ABS-201 - The study will start with a single ascending dose (SAD) followed by a multiple ascending dose (MAD) study, with endpoints focused on hair density and terminal hair count [63][65] Question: Interest in large pharma partnerships and therapeutic areas - The focus is broad, with particular interest in immunology and oncology, leveraging the company's de novo design capabilities [67][70] Question: Expectations for interim readout of ABS-101 - The interim readout is expected to demonstrate sustained elevation of soluble TL1A in serum after a single dose [83][84] Question: Progress on the HIV program with Caltech - The collaboration is funded by the Gates Foundation, with plans to develop the molecule further based on preclinical data [98] Question: 2025 spending ramp and cash burn rate - The company reiterated that its balance sheet will support operations into the first half of 2027, with plans to advance Phase I studies for ABS-101 and ABS-201 [102]

Core Insights - Absci Corporation has made significant advancements in its AI Integrated Drug Creation platform, showcasing new breakthroughs and preparing to enter clinical stages with its drug candidates [2][5][6] Financial Performance - For Q4 2024, Absci reported revenue of $0.7 million, an increase from $0.3 million in Q4 2023 [8] - Full-year revenue for 2024 was $4.5 million, down from $5.7 million in 2023 [11] - Research and development expenses rose to $18.4 million in Q4 2024 from $12.3 million in Q4 2023, totaling $63.9 million for the full year, up from $48.1 million in 2023 [9][12] - Selling, general, and administrative expenses decreased to $8.8 million in Q4 2024 from $9.3 million in Q4 2023, totaling $36.2 million for the full year, down from $37.8 million in 2023 [10][13] - The net loss for Q4 2024 was $29.0 million, compared to $23.5 million in Q4 2023, with a full-year net loss of $103.1 million, an improvement from $110.6 million in 2023 [10][13] Pipeline Developments - The company is advancing its internal pipeline, with ABS-101 expected to initiate Phase 1 studies in the coming months, marking a significant milestone [2][5] - ABS-201, targeting androgenic alopecia, has shown promising preclinical results and is anticipated to enter Phase 1 clinical trials in early 2026 [5][6] - ABS-301, a fully human antibody for an undisclosed immuno-oncology target, has completed initial in vivo target validation studies, supporting its further development [6] Strategic Collaborations - Absci entered a strategic collaboration with AMD, which includes a $20 million investment to enhance AI drug discovery capabilities [5] - New partnerships have been established with Owkin and Invetx to co-develop therapeutic candidates and leverage Absci's generative AI models [5] Cash Position - As of December 31, 2024, Absci had cash, cash equivalents, and short-term investments totaling $112.4 million, sufficient to fund operations into the first half of 2027 [7][14]