AstraZeneca Buys Modella AI: Why Biomarkers, Not Molecules, Are the Real Prize

来源：Onco daily | 发布时间：2026-01-30

摘要：2026 年 1 月，阿斯利康在 J.P. Morgan 医疗保健大会上宣布收购波士顿肿瘤领域 AI 初创公司 Modella AI（系 2025 年 7 月双方合作的升级），将其生成式 AI 与智能体 AI 工具全面嵌入内部肿瘤研发体系，核心聚焦临床试验加速、生物标志物发现及患者分层优化，以解决肿瘤药因患者选择不当导致的研发失败痛点，此举标志着大型药企已从 AI 合作试点转向 AI 平台自主化，凸显肿瘤领域作为生物制药 AI 应用主战场的地位，未来 1-3 年有望推动临床试验革新、AI 驱动伴随诊断突破，并引发行业同行的 AI 研发竞争升级。

阿斯利康收购Modella AI：为何生物标志物而非分子才是真正的收获

AstraZeneca, one of the world’s pharmaceutical giants with a major oncology franchise, is doubling down on artificial intelligence to sharpen its edge in cancer R&D. In a landmark move announced at the J.P. Morgan Healthcare Conference 2026, AstraZeneca acquired Modella AI, a Boston-based biomedical AI startup focused on oncology. The deal expands a multi-year partnership (inked in July 2025) into a full acquisition, signaling AstraZeneca’s intent to embed AI into oncology drug development rather than just collaborate at arm’s length. Financial details were not disclosed, but the strategic value is clear: this is an acquisition aimed at accelerating oncology clinical trials, uncovering new biomarkers, and improving patient stratification across AstraZeneca’s cancer pipeline.阿斯利康（AstraZeneca）是全球制药巨头之一，拥有庞大的肿瘤学业务，该公司正加大对人工智能的投入，以提升其在癌症研发领域的优势。在2026年摩根大通医疗健康大会（J.P. Morgan Healthcare Conference）上，阿斯利康宣布了一项具有里程碑意义的举措——收购Modella AI。Modella AI是一家总部位于波士顿的生物医学人工智能初创公司，专注于肿瘤学领域。此次收购将双方于2025年7月达成的多年合作关系升级为全面收购，这表明阿斯利康打算将人工智能融入肿瘤药物研发中，而不仅仅是进行远距离合作。交易的财务细节尚未披露，但其战略价值显而易见：此次收购旨在加快肿瘤学临床试验进程、发现新的生物标志物，并改善阿斯利康整个癌症研发管线中的患者分层。

Deal Structure: From Partnership to Acquisition交易结构：从合作到收购

The AstraZeneca–Modella AI relationship began as a partnership and quickly proved its value. In July 2025, the two entered a multi-year collaboration to deploy Modella’s AI models on AstraZeneca’s trials, aiming to speed up timelines and uncover hidden biomarkers in ongoing studies. That pilot was evidently successful. Now, less than a year later, AstraZeneca has moved to bring Modella fully in-house via acquisition. A clear sign that the pharma is confident in the technology. Key elements of the deal include:阿斯利康与Modella AI的合作始于一次伙伴关系，其价值很快得到了证明。2025年7月，双方达成了一项多年合作协议，将Modella的人工智能模型应用于阿斯利康的试验中，旨在加快试验进度，并在进行中的研究中发现隐藏的生物标志物。那次试点显然是成功的。如今，不到一年之后，阿斯利康已通过收购将Modella完全纳入旗下。这一明确迹象表明，这家制药公司对该技术充满信心。此次交易的关键内容包括：

• Full integration of Modella’s platform: Modella’s generative AI and agentic AI tools will be embedded directly into AstraZeneca’s oncology R&D organization. Instead of a vendor-client relationship, Modella’s team and tech become an internal engine for AstraZeneca.Modella平台的完全整合：Modella的生成式人工智能和智能体人工智能工具将直接嵌入阿斯利康的肿瘤学研发机构。Modella的团队和技术不再是供应商与客户的关系，而是成为阿斯利康的内部引擎。
• Focus on clinical development and biomarkers: The stated goals are to accelerate clinical trials, enhance biomarker discovery, and drive data-driven decision-making in AstraZeneca’s oncology portfolio. This suggests Modella’s AI will be applied to ongoing trial datasets to identify which patients benefit most, which biomarkers predict response, etc.专注于临床开发和生物标志物：既定目标是加快临床试验、加强生物标志物的发现，并在阿斯利康的肿瘤学产品组合中推动数据驱动的决策。这表明Modella的人工智能将应用于正在进行的试验数据集，以确定哪些患者受益最大、哪些生物标志物可预测反应等。
• No financial terms disclosed: AstraZeneca and Modella have not released the price tag or deal terms. The move is less about near-term financials and more about long-term capability-building. By acquiring Modella AI, AstraZeneca secures exclusive access to its AI platform (and talent) for competitive advantage in oncology.未披露任何财务条款：阿斯利康和Modella尚未公布价格或交易条款。此举与其说是为了短期财务状况，不如说是为了长期能力建设。通过收购Modella AI，阿斯利康获得了其人工智能平台（及人才）的独家使用权，以在肿瘤学领域获得竞争优势。

This acquisition structure (vs. extended partnership) underscores a broader industry trend: big pharma is shifting from experimenting with AI vendors to owning AI platforms outright. AstraZeneca now joins the likes of Roche and others who have made bold plays to internalize key digital assets (e.g. Roche’s acquisitions of Flatiron Health and Foundation Medicine in data/diagnostics).这种收购结构（而非长期合作）凸显了一个更广泛的行业趋势：大型制药公司正从与人工智能供应商合作试验转向直接拥有人工智能平台。阿斯利康现在加入了罗氏等公司的行列，这些公司大胆地将关键数字资产内部化（例如，罗氏在数据/诊断领域收购了Flatiron Health和Foundation Medicine）。

Why This Deal Matters for AstraZeneca and Industry Rivals这笔交易为何对阿斯利康及其行业竞争对手至关重要

In the unforgiving arena of oncology drug development, late-stage trial failures and narrow therapeutic windows are constant risks. AstraZeneca’s purchase of Modella AI is strategically aimed at de-risking R&D and boosting precision oncology efforts:在残酷的肿瘤药物研发领域，后期试验失败和狭窄的治疗窗口是持续存在的风险。阿斯利康收购Modella AI，其战略目标是降低研发风险并加强精准肿瘤学方面的努力：

• R&D Productivity and Trial Acceleration: AstraZeneca’s oncology pipeline is extensive, with multiple drugs in development for lung, breast, and other cancers. By leveraging Modella’s AI on trial data, AstraZeneca hopes to shorten time-to-signal, i.e. detect early whether a drug is working and in which subgroup of patients. AI-driven trial acceleration can save hundreds of millions by flagging failures faster or guiding adaptive trial designs.研发生产力与试验加速：阿斯利康的肿瘤学研发管线十分广泛，有多种药物正在研发中，用于治疗肺癌、乳腺癌及其他癌症。通过将Modella的人工智能应用于试验数据，阿斯利康希望缩短信号出现时间，即尽早检测出药物是否有效以及对哪些患者亚群有效。人工智能驱动的试验加速能够通过更快地标记失败案例或指导适应性试验设计，节省数亿美元成本。
• Biomarker-Driven Precision Oncology: The real prize is biomarkers. Modern cancer drugs often succeed or fail based on finding the right patient population. Modella’s AI excels at analyzing pathology and genomic data to find patterns. AstraZeneca can now systematically hunt for such biomarkers with AI, improving its chances of late-stage trial success and regulatory approval for targeted therapies.生物标志物驱动的精准肿瘤学：真正的关键在于生物标志物。现代抗癌药物的成败往往取决于能否找到合适的患者群体。Modella的人工智能擅长分析病理学和基因组数据以发现规律。阿斯利康现在可以借助人工智能系统地寻找此类生物标志物，从而提高其后期试验成功以及靶向疗法获得监管批准的几率。
• Competitive Pressure on Big Pharma: This move has competitive ripples. Pharma peers like Pfizer, Novartis, Merck, Bristol Myers Squibb (BMS), and Roche are all pursuing AI strategies, but few have outright acquired an AI company for R&D. AstraZeneca may now leap ahead in having an integrated AI capability. 大型制药公司面临的竞争压力：这一举措引发了连锁竞争反应。辉瑞、诺华、默克、百时美施贵宝（BMS）和罗氏等制药同行都在推行人工智能战略，但很少有公司直接收购一家人工智能公司用于研发。阿斯利康现在可能在整合人工智能能力方面实现跨越式领先。
• From AI Hype to Real Platforms: The acquisition also signals a turning point in pharma’s adoption of AI. A few years ago, “AI in drug discovery” was largely hype. Lots of press releases, fewer tangible results. Many deals were transactional or pilot projects. By contrast, AstraZeneca bringing Modella in-house indicates a belief that AI is now essential infrastructure for drug development. It’s not about flashy demos; it’s about day-to-day use of AI in trial design, diagnostics, and decision-making.从人工智能炒作到真正的平台：这一收购也标志着制药行业采用人工智能的转折点。几年前，“人工智能在药物研发中的应用”在很大程度上只是炒作。相关的新闻发布会不少，但切实的成果却较少。许多交易都是一次性的或试点项目。相比之下，阿斯利康将Modella收归旗下表明，该公司认为人工智能现已成为药物研发的重要基础设施。这无关花哨的演示，而是关乎在试验设计、诊断和决策过程中日常使用人工智能。

Oncology: The Primary Battleground for AI肿瘤学：人工智能的主要战场

It’s no coincidence this transformative AI deal centers on cancer. Oncology is the primary battleground for AI in biopharma for several reasons:这项具有变革意义的人工智能交易以癌症为核心并非巧合。出于几个原因，肿瘤学是生物制药领域人工智能的主要战场：

• Data Complexity & Volume: Cancer research produces vast multimodal datasets. Pathology images, genomic sequences, longitudinal patient outcomes, etc. This richness is a goldmine for AI, which thrives on big data. AI can sift through these layers of data far faster than humans. For instance, analyzing a whole-slide pathology image manually might take a pathologist hours, whereas an AI model can quantify every cell and correlate it with genomic markers in minutes. The payoff is discovering non-obvious patterns, e.g., an AI might learn that a certain immune-cell spatial pattern in a tumor predicts response to a drug, enabling a new predictive biomarker.数据复杂性与体量：癌症研究产生了海量的多模态数据集，包括病理图像、基因组序列、患者的纵向结局等。这种丰富性对人工智能来说是一座金矿，因为人工智能在大数据环境中能大显身手。人工智能筛选这些层级数据的速度远超人类。例如，病理学家手动分析一张全切片病理图像可能需要数小时，而人工智能模型能在几分钟内量化每个细胞，并将其与基因组标记物关联起来。其成果是发现了不明显的模式，例如，人工智能可能会发现肿瘤中某种免疫细胞的空间模式可预测对药物的反应，从而形成一种新的预测性生物标志物。
• High Unmet Need & Reward: Oncology has countless subtypes of disease and many patients who still lack effective treatments. Any edge in trial success or patient selection can translate to lives saved (and significant revenue). This urgency makes oncology teams more willing to adopt new tech. It’s telling that most pharma–AI collaborations in recent years have focused on cancer. Simply put, if AI can increase the success rate of cancer trials, it’s enormously valuable. That appears to be Modella’s sweet spot, not inventing new molecules from scratch, but making sure the right treatments get to the right patients.高度未被满足的需求与回报：肿瘤学领域有无数种疾病亚型，许多患者仍然缺乏有效的治疗方法。临床试验成功或患者筛选方面的任何优势都可能转化为拯救生命（以及可观的收入）。这种紧迫性使得肿瘤学团队更愿意采用新技术。值得注意的是，近年来大多数制药公司与人工智能的合作都集中在癌症领域。简而言之，如果人工智能能够提高癌症试验的成功率，它将具有巨大的价值。这似乎正是Modella的优势所在——并非从零开始发明新分子，而是确保合适的治疗方法用于合适的患者。
• Biomarker Discovery vs. Molecule Discovery: The AstraZeneca–Modella deal highlights that AI’s hottest role now is accelerating biomarker discovery rather than molecule discovery. A few years ago, headlines touted AI designing new drugs (which is progressing, but will take time to prove). Meanwhile, pharma learned the hard way that a breakthrough drug still fails if tested in the wrong patients. The trend now is to use AI to shorten time-to-signal, not just time-to-clinic.生物标志物发现与分子发现：阿斯利康与Modella的交易凸显出，人工智能目前最热门的作用是加速生物标志物的发现，而非分子发现。几年前，新闻头条大肆宣扬人工智能设计新药（这一领域正在取得进展，但还需要时间来证明）。与此同时，制药行业深刻认识到，如果在错误的患者身上进行试验，即使是突破性的药物也会失败。现在的趋势是利用人工智能缩短信号出现的时间，而不仅仅是缩短进入临床的时间。
• Ecosystem of Oncology AI Startups: Modella AI is part of a burgeoning landscape of AI-in-oncology ventures. This includes companies focusing on digital pathology (like PathAI, Paige), clinical trial data mining (like Owkin, which partners with BMS), and multi-omics analysis (like Tempus or Foundation Medicine’s AI efforts). By acquiring Modella, AstraZeneca obtains a unique combo: a digital pathology powerhouse plus a platform for multi-omics and trial data integration. It sets a precedent, we might see other big pharmas eyeing similar startups. Oncology is where these chess moves are happening first, because the field’s complexity makes AI assistance not just nice-to-have, but increasingly essential.肿瘤学人工智能初创企业生态系统：Modella AI 是蓬勃发展的肿瘤学人工智能企业领域的一部分。这其中包括专注于数字病理学的公司（如 PathAI、Paige）、专注于临床试验数据挖掘的公司（如与百时美施贵宝合作的 Owkin）以及专注于多组学分析的公司（如 Tempus 或 Foundation Medicine 的人工智能项目）。通过收购 Modella，阿斯利康获得了一个独特的组合：一个数字病理学巨头，以及一个多组学和试验数据整合平台。这开创了一个先例，我们可能会看到其他大型制药公司也在关注类似的初创企业。肿瘤学领域是这些战略举措首先展开的地方，因为该领域的复杂性使得人工智能辅助不仅是一种锦上添花，而且正变得越来越不可或缺。