摘要：Advancing drug discovery. Helping treatments reach patients faster. Diversifying clinical trials. Here’s a look at how Johnson & Johnson is harnessing AI to help create a healthier world.
In a time when chronic diseases are on the rise and people are living longer than ever before, novel solutions for better patient care are urgently needed. In healthcare’s next chapter, a new type of technology will play a bigger role than ever before.
Enter artificial intelligence, or AI. Rooted in the simulation of human intelligence by computer systems and machines, AI has the potential to transform how humans learn, work and interact with one another in every aspect of life.
It’s also primed to revolutionize healthcare.
“The rapid growth in available healthcare-related data in recent years allows us to ask bigger questions,” says Jeff Headd, Vice President, Commercial Data Science, Janssen North America Business Technology. “Using the latest innovations in AI and machine learning (ML), we are able to quickly analyze these vast datasets (including electronic medical records, lab results or even medical imaging like X-rays, MRIs and CT scans), uncover new insights and then drive actions with real potential to improve patient outcomes.”
The promise that AI holds is why Johnson & Johnson is actively using the technology in different ways, from speeding up the process of discovering new medicines to helping surgeons analyze the results of procedures. It’s also why, during this year’s South by Southwest conference, the company hosted a panel about AI’s role in transforming healthcare.
“There’s a deep demand for solutions in the healthcare space,” says Shan Jegatheeswaran, Global Head of MedTech Digital, Johnson & Johnson, who spoke on the panel. “But it’s important to remember that the most sophisticated thing in the clinical workflow is still the human brain. The role of AI is to augment a human decision or action in a way that improves speed, quality or both.”
We’re taking a closer look at five ways AI is helping drive healthcare forward—and how Johnson & Johnson is using it to help improve the quality of medical care around the world.
Facilitating earlier detection of diseas
When it comes to catching and diagnosing diseases earlier, AI can be a real game changer. By applying AI to data derived from or generated by common diagnostic tests, such as electrocardiograms and echocardiograms, providers could diagnose diseases more accurately, prevent delays in care and potentially save lives.
Take pulmonary hypertension (PH) and cardiac amyloidosis, two progressive and often-fatal diseases. Despite the existence of treatments, both diseases are commonly misdiagnosed early on, given that their respective symptoms mimic those of other, more common diseases.
To help diagnose these diseases, Johnson & Johnson has teamed up with collaborators Anumana and Mayo Clinic for pulmonary hypertension, and Ultromics Ltd. and Atman Health for cardiac amyloidosis to develop AI algorithms aimed at helping detect these diseases early on.
“Our goal has been to develop AI-enhanced tools that can be seamlessly integrated into the current clinical workflow of physicians,” says Mona Selej, Senior Director, Cardiovascular-Metabolism and Pulmonary Hypertension, Data Science & Digital Health, R&D, Janssen Pharmaceutical Companies of Johnson & Johnson and a PH physician by training. “We determined which tests patients commonly receive early in their journeys to diagnosis—electrocardiograms in the case of PH and echocardiograms in the case of cardiac amyloidosis—and developed AI algorithms that could detect subtleties that are invisible to the naked eye, suggesting patients should be flagged for confirmatory testing.”
The PH algorithm and Ultromics’ cardiac amyloidosis algorithm have both received Breakthrough Device Designation from the U.S. Food & Drug Administration and, if approved, could help facilitate earlier, more accurate diagnoses, leading to patients receiving treatment sooner.
Beyond early detection algorithms, medical devices, including connected devices, robotic platforms and digital solutions are also evolving with AI to enhance their capabilities.
Johnson & Johnson MedTech’s Monarch™ Platform for bronchoscopy, for example, lets physicians examine areas of the lung that are more difficult to access with conventional bronchoscopes—and that can aid in earlier lung cancer diagnosis. The flexible robotics system uses preoperative CT scans of the lungs to inform the procedure, but tracking objects in such a dynamic environment in real time can be complex. The Monarch R&D team uses AI and ML algorithms to develop and refine the Monarch Platform’s navigation, which helps physicians guide the bronchoscope during lung biopsy procedures and allows them to locate a potential tumor more accurately. This leads to more accurate diagnosis and treatment.
Driving drug discovery
Traditionally, discovering and developing new drugs to treat disease is a long and complex undertaking, but AI is primed to help accelerate this process.
To develop medicines, researchers need to understand what biological and genetic variations cause diseases to develop. By applying AI to anonymized medical datasets, such as electronic health records or lab results, scientists can fill in missing information as to what causes those diseases.
AI is also enabling researchers to develop more targeted medicines, driving progress toward precision medicine.
For example, in oncology an AI algorithm can be applied to digitized images of biopsies to help identify subtle differences between tumors, pointing to the presence of genetic mutations in a subset of patients. Researchers can use these findings to develop medicines specifically designed for that subset of patients. Those same algorithms that can help identify genetic mutations could then be used to find these patients in the real world to facilitate clinical trial recruitment and clinical decision-making.
“Drug discovery is an extremely challenging process with only a small percentage of lead compounds moving into clinical trials and an even smaller percentage becoming approved medicines,” says Chris Moy, Scientific Director, Oncology, Data Science & Digital Health, R&D, Janssen. “AI is not only helping us identify the right targets for complex diseases, but it’s also helping us design fit-for-purpose molecules to treat diseases and optimize them to provide targeted treatment to the disease while also reducing the impact of side effects.”
Together, these applications of AI will help researchers place the most promising candidate drugs into clinical development, with the ultimate goal of improving the probability of successfully bringing a drug to market and rapidly getting new treatments to patients who need them the most.
Enabling more targeted clinical trial recruitment
One of the biggest challenges when it comes to running clinical trials is quickly and efficiently recruiting and enrolling patients that meet the selection criteria. Adopting AI technology into the process may help solve this problem.
Example: At Johnson & Johnson, researchers are applying AI and ML algorithms to large anonymized datasets to identify and locate clinical research sites with patients who could potentially benefit from medicines being studied. The clinical trial operations team can then work to determine the likelihood of enrolling the newly identified sites into their trials.
Our goal is to leverage the power of AI to bring trials to more patients, rather than waiting for patients to come to us.
Nicole Turner, Senior Director of Global Development, Data Science & Digital Health, R&D, Janssen
“Historically, many clinical trials have largely taken place at major academic medical centers, but we know that not all patients have access to these centers,” says Nicole Turner, Senior Director of Global Development, Data Science & Digital Health, R&D, Janssen. “Our goal is to leverage the power of AI to bring trials to more patients, rather than waiting for patients to come to us.”
Data and AI are also helping researchers diversify clinical trials, as advanced analytics are finding locations and healthcare institutions where diverse patients are more likely to be treated. Researchers can then prioritize recruiting eligible patients from those study sites into clinical trials. This is critical, given the importance of ensuring medicines are studied in diverse patient populations representative of those impacted by diseases.
Ensuring treatments can reach patients
First comes the discovery and development of therapies, medications and other healthcare products. Then, the next critical juncture is making sure these products reach patients, and AI can help with that, too.
Stocking products in the hospitals, pharmacies, clinics and other healthcare facilities where they’re needed requires an accurate prediction of supply and demand. This can be challenging, as a wide range of factors can impact the supply chain, including market trends, economic disruptions, supplier issues and more.
AI-powered solutions can help prioritize which locations will be greatly impacted in order to quickly respond to risk factors that may otherwise affect the ability to deliver products to people who count on them.
“It’s our responsibility to make sure patients and customers have reliable access to the transformational therapies our company creates,” says Vishal Varma, Director, Supply Chain Digital & Data Science and Operations Research, Johnson & Johnson Services, Inc. “AI is helping us build a stable and resilient supply chain so we can deliver on that obligation.”
For example, Johnson & Johnson is using advanced ML algorithms to sift through and analyze large amounts of data, including demand fluctuations and supplier performance, as well as to help predict the impact of real-time events that may disrupt the supply chain (think severe weather events and economic disruptions). AI-powered solutions can help prioritize which locations will be greatly impacted in order to quickly respond to risk factors that may otherwise affect the ability to deliver products to people who count on them.
Ensuring patients have access to the right products also requires timely communication with healthcare providers.
“Another way Johnson & Johnson is using ML is to improve our understanding of disease progression, which allows us to anticipate when a patient may benefit from one of our medicines,” says Headd.
This application is part of a Johnson & Johnson global capability known as Engagement.ai, designed to guide the company’s engagement with healthcare professionals. Engagement.ai is powered by AI and ML models trained on extensive datasets to provide insights that maximize the company’s ability to support providers and their patients.
“These insights from Engagement.ai allow us to prioritize when, where and how we connect with healthcare providers to ensure they have relevant and appropriate information when making treatment decisions,” says Headd.
Analyzing the OR for efficiency and physician learning
In the operating room (OR), surgical video is frequently taken during procedures to provide education, research methodologies and quality improvement strategies to medical professionals.
Johnson & Johnson MedTech is evolving a portfolio of digital solutions for the OR that use AI algorithms to essentially “cut a highlight reel of these videos” in a matter of minutes, Jegatheeswaran explains, so surgeons can re-watch significant events from their procedures. Without AI, this process could take hours—even days—to complete.
“Surgeons are a lot like high-performance athletes,” says Jegatheeswaran. “New and learning surgeons want to see how they performed and learn from their performances and how others performed. But no one wants to sit and watch hours of footage from the full procedure.”
Now, surgeons can look at what happened during procedures practically in real time and share the video with residents and peers, offering valuable post-case analysis and learning opportunities.
“Once you get enough of these enriched surgical videos, you can start running algorithms on the behaviors, tactics and movements that create positive and negative outcomes during surgery,” Jegatheeswaran says.
Surgical video is one of many potential applications of AI in the OR, according to Larry Jones, Chief Information Officer, Johnson & Johnson MedTech.
“Our vision in Johnson & Johnson MedTech is to make healthcare smarter, less invasive, more personalized and more connected,” Jones says. “By combining a wealth of data stemming from surgical procedures and increasingly sophisticated AI technologies, we can transform the experience of patients, doctors and hospitals alike.”
Ultimately, that’s what AI in healthcare is all about: better serving providers and patients. “When we use AI, it’s always with a purpose,” Jegatheeswaran says. “Our Credo states that our patients and customers come first, and that will continue as we move forward with this technology.”
随着人类寿命不断增加、慢性疾病不断增多，迫切需要新的解决方案来改善病人护理现状。在医疗保健的下一个篇章中，人工智能将发挥前所未有的作用。
杨森（Janssen）北美业务技术部商业数据科学副总裁Jeff Headd说：“近年来，可用的医疗保健相关数据迅速增长，这使我们能够提出更多问题。利用人工智能和机器学习（ML）方面的最新创新，我们能够快速分析这些庞大的数据集（包括电子病历、实验室结果甚至X光、核磁共振成像和CT扫描等医学影像），发现新的见解，然后推动行动，真正改善患者的治疗效果。”
强生医疗技术数字部门全球主管Shan Jegatheeswaran说：“医疗保健领域对解决方案的需求很高。但是要记住，临床工作流程中最复杂的东西仍然是人脑。人工智能的作用是增强人类的决策或行动，从而提高速度、质量或两者兼而有之。”
我们将进一步了解人工智能推动医疗保健发展的五种方式，以及强生如何利用人工智能帮助提高全球医疗保健质量。
便于更早地发现疾病
在更早发现和诊断疾病方面，人工智能可以真正改变游戏规则。通过将人工智能应用于心电图和超声心动图等常见诊断测试所衍生或生成的数据，医疗服务提供者可以更准确地诊断疾病，防止护理延误，并有可能挽救生命。
以肺动脉高压（PH）和心脏淀粉样变性为例，这是两种进展性疾病，通常会致命。尽管这两种疾病都有治疗方法，但由于它们各自的症状与其他更常见的疾病相似，因此在早期常被误诊。为了帮助诊断这些疾病，强生公司与合作者Anumana和Mayo Clinic合作治疗肺动脉高压，与Ultromics和Atman Health合作治疗心脏淀粉样变性，通过开发AI算法，帮助及早发现这些疾病。强生旗下杨森制药研发部数据科学与数字健康、心血管代谢与肺动脉高压高级总监Mona Selej说：“我们的目标是开发可无缝集成到医生当前临床工作流程中的人工智能增强工具。我们确定了患者在诊断早期通常会接受哪些检查–PH病例中的心电图和心脏淀粉样变性病例中的超声心动图–并开发了人工智能算法，可以检测出肉眼看不到的细微差别，提示患者应进行标记以进行确诊检查。”PH算法和Ultromics的心脏淀粉样变性算法都获得了FDA的突破性设备认定，如果获得批准，将有助于更早、更准确地做出诊断，使患者更快接受治疗。除早期检测算法外，医疗设备（包括联网设备、机器人平台和数字解决方案）也在不断发展人工智能，以增强其功能。
例如，强生医用于支气管镜检查的Monarch™ 平台可让医生检查传统支气管镜较难进入的肺部区域，从而有助于更早地诊断肺癌。灵活的机器人系统使用术前肺部CT扫描为手术提供信息，但在这样一个动态环境中实时跟踪物体可能非常复杂。Monarch研发团队使用人工智能和ML算法来开发和完善Monarch平台的导航功能，帮助医生在肺部活检手术中引导支气管镜，使他们能够更准确地定位潜在肿瘤。这将带来更准确的诊断和治疗。
推动药物研发
传统上，发现和开发治疗疾病的新药是一项漫长而复杂的工作，但人工智能可以帮助加快这一进程。
为了开发药物，研究人员需要了解哪些生物和基因变异会导致疾病的发生。通过将人工智能应用于匿名医疗数据集（如电子健康记录或实验室结果），科学家们可以填补缺失的信息，了解导致这些疾病的原因。
人工智能还能让研究人员开发出更有针对性的药物，推动精准医疗的发展。例如，在肿瘤学领域，人工智能算法可应用于活检的数字化图像，帮助识别肿瘤之间的细微差别，从而发现部分患者存在基因突变。研究人员可以利用这些发现，开发专为这部分患者设计的药物。这些有助于识别基因突变的算法也可用于在现实世界中寻找这些患者，以促进临床试验招募和临床决策。杨森公司研发部肿瘤学、数据科学与数字健康科学总监Chris Moy说：“药物发现是一个极具挑战性的过程，只有一小部分先导化合物进入临床试验，而成为批准药物的比例则更小。人工智能不仅能帮助我们为复杂的疾病确定正确的靶点，还能帮助我们设计出适合治疗疾病的分子，并对其进行优化，以提供针对疾病的靶向治疗，同时减少副作用的影响。”
人工智能的这些应用将共同帮助研究人员将最有希望的候选药物投入临床开发，最终目标是提高药物成功上市的概率，并迅速将新疗法提供给最需要的患者。
实现更有针对性的临床试验招募
开展临床试验的最大挑战之一是快速高效地招募和注册符合选择标准的患者。在这一过程中采用人工智能技术可能有助于解决这一问题。
例如，在强生，研究人员正在将人工智能和ML算法应用于大型匿名数据集，以识别和定位拥有相应患者（可能从正在研究的药物中获益）的临床研究机构。然后，临床试验运营团队可以确定将新确定的研究机构纳入其试验的可能性。 杨森研发部数据科学与数字健康全球开发高级总监Nicole Turner说：“从历史上看，许多临床试验主要在大型学术医疗中心进行，但我们知道，并非所有患者都有机会前往这些中心。我们的目标是利用人工智能的力量，将试验带给更多患者，而不是等待患者来找我们。”
数据和人工智能也在帮助研究人员实现临床试验的多样化，因为先进的分析技术正在发现不同患者更有可能接受治疗的中心和医疗机构。然后，研究人员可以优先从这些研究地点招募符合条件的患者参与临床试验。这一点至关重要，因为必须确保在代表受疾病影响的不同患者群体中对药物进行研究。
确保患者获得治疗
首先是发现和开发治疗方法、药物和其他保健产品。然后，下一个关键时刻就是确保这些产品送达患者手中，而人工智能也能在这方面提供帮助。
要在医院、药房、诊所和其他医疗机构储备所需的产品，就必须准确预测供需情况。这可能具有挑战性，因为影响供应链的因素有很多，包括市场趋势、经济混乱、供应商问题等。强生供应链数字与数据科学及运营研究总监Vishal Varma表示：“我们有责任确保患者和客户能够可靠地获得我们公司创造的变革性疗法。人工智能正在帮助我们建立一个稳定而有弹性的供应链，以便我们能够履行这一义务。”例如，强生正在使用先进的ML算法来筛选和分析大量数据，包括需求波动和供应商绩效，以及帮助预测可能扰乱供应链的实时事件（如恶劣天气事件和经济中断）的影响。人工智能驱动的解决方案可以帮助确定哪些地点将受到重大影响的优先次序，以便快速应对可能影响向人们提供产品的能力的风险因素。确保患者能够获得正确的产品还需要与医疗保健提供商及时沟通。Headd说：“强生使用ML的另一种方法是提高我们对疾病进展的了解，这使我们能够预测患者何时可能从我们的药物中获益。Engagement.ai提供的这些洞察力使我们能够优先考虑何时、何地以及如何与医疗保健提供商建立联系，以确保他们在做出治疗决定时获得相关的适当信息。”
该应用是强生名为Engagement.ai的全球能力的一部分，旨在指导公司与医疗保健专业人员的合作。Engagement.ai由在大量数据集上训练的人工智能和 ML 模型提供支持，能够提供洞察力，最大限度地提高公司为医疗服务提供者及其患者提供支持的能力。
在手术室 (OR) 中，手术过程中经常会拍摄手术视频，以便为医疗专业人员提供教育、研究方法和质量改进策略。
Jegatheeswaran解释说：”强生医疗正在为手术室开发一系列数字解决方案，利用人工智能算法在几分钟内‘剪辑出这些视频的精彩片段’，这样外科医生就可以重新观看手术过程中的重要事件。如果没有人工智能，这一过程可能需要数小时甚至数天才能完成。外科医生很像高水平运动员，新外科医生和正在学习的外科医生都希望看到自己的表现，并从自己和他人的表现中学习。但没有人愿意坐下来观看数小时的完整手术录像。现在，外科医生可以实时查看手术过程中发生的情况，并与住院医生和同行分享视频，提供宝贵的术后分析和学习机会。一旦你获得了足够多的这些丰富的手术视频，你就可以开始对手术过程中产生积极和消极结果的行为、战术和动作进行算法运算。”强生医疗首席信息官Larry Jones 表示：“手术视频是人工智能在手术室的众多潜在应用之一。”Jones说：“强生医疗的愿景是让医疗保健更智能、更微创、更个性化和更互联。通过将手术过程中产生的大量数据与日益成熟的人工智能技术相结合，我们可以改变患者、医生和医院的就医体验。”Jegatheeswaran说：“归根结底，这就是人工智能在医疗保健领域的意义所在：更好地服务提供者和患者。当我们使用人工智能时，总是有目的的。我们的信条规定，我们的患者和客户是第一位的，这一点将在我们推进这项技术的过程中继续保持下去。”