The “Work-Life Integration” Philosophy: Passion as Fuel

Early in the podcast, Ashis tackled the traditional concept of “work-life balance” with a refreshing perspective, advocating instead for “work-life integration”. Drawing inspiration from books like “Conscious Business,” he suggests that when work is driven by passion, the lines between personal and professional life naturally blur. For Ashis, his work is a source of energy and passion, much like eating is an intrinsic part of life, not a separate task to be balanced. This deep personal investment, he believes, is evident in his fervent discussions about digital health, demonstrating how he truly enjoys the space. This approach underscores that true fulfilment comes from aligning one’s work with their core passions, allowing for a more harmonious and energetic existence, rather than a constant struggle for an elusive balance.

The Physician-Technologist Hybrid: Bridging the Gap

Ashis’s journey from a paediatric gastroenterologist who never stopped practicing to a Chief Digital Technology Officer is a defining aspect of his approach. He emphatically states, “I’m absolutely a doctor first, a technologist second, and I am to this day.” This unique dual perspective is a key differentiator that he believes brings immense value to discussions and solutions in healthcare technology. He intimately understands the frustrations faced by frontline clinicians when “logical” technology solutions, conceived by the C-suite, inadvertently add burden to their workflows. His commitment to spending time with clinical teams, observing their daily realities, and truly understanding their problems from the ground up ensures that the technology solutions implemented are not just theoretically sound but actually solve real problems without creating new friction. This commitment to bridging the gap between clinical practice and technological innovation is crucial for effective digital transformation.

The Driving Force: Democratizing Access to Right Care

A deeply personal experience from Ashis’s childhood fundamentally shaped his mission to democratise healthcare access. At eight years old, during a family trip to rural India, he contracted typhoid fever, which was initially misdiagnosed as malaria. Severely ill and rapidly losing weight, his life was saved by a physician cousin who correctly identified and treated his condition. This profound experience of receiving the “wrong care” until he gained access to the “right care” ignited his passion. He questioned, “How do we give, how do we distribute? How do we democratize? How do we get the right care to all people?” This foundational belief continues to fuel his digital transformation efforts, aiming to leverage technology not just for efficiency but to ensure equitable access and better health outcomes for everyone.

Agentic AI: The Workflow Orchestrator of the Future

Ashis suggests that the two most critical discussions in healthcare today are “agentic AI and change management”. He is a self-proclaimed “techno optimist” but also a pragmatist, wanting technology that genuinely works and solves problems. His excitement for Agentic AI stems from its potential as a “workflow orchestrator,” a capability largely missing in current point solutions or even the Electronic Medical Record (EMR) which, despite its utility, can be burdened by “friction and lots of clicks”. Healthcare, he argues, operates in complex workflows, not isolated moments. He states – “Healthcare is about workflows. Healthcare isn’t about a moment in time.” He further notes, “The only two things that we should be talking about in healthcare right now is Agentic AI and change management.“

Every “handoff” in a patient’s journey – from finding care, to scheduling, receiving treatment, and post-care – presents opportunities for friction and significant waste due to a lack of coordination across vertically structured hospital systems. Agentic AI, by orchestrating across these traditionally siloed operations, promises to improve patient experience, enhance coordination, improve outcomes, and ultimately reduce costs by eliminating the “white space” between different care episodes.

HSS’s “Focus Factory” Advantage: A Lighthouse for Innovation

Ashis chose to join the Hospital for Special Surgery (HSS) for a very purposeful reason, despite having worked for much larger organisations. He refers to HSS as a “focus factory,” dedicated exclusively to musculoskeletal care. This specialisation, while seemingly narrow, actually impacts a significant portion of the population (30-40% experience mobility problems) and involves many algorithmic and elective procedures, making it an ideal environment for the application of Agentic AI. Unlike larger, more diverse healthcare systems where orchestrating across multiple complex specialties (e.g., cardiac, cancer) would take “5 to 10, 20 years,” HSS’s singular focus allows for deep vertical development and a much shorter timeline of “two to five years” for implementing comprehensive digital transformation. Ashis envisions HSS becoming a “lighthouse” for healthcare, demonstrating the feasibility of automating backend processes and orchestrating care workflows. The ambition is not only to show what’s possible but also to codify HSS’s world-class knowledge and distribute it globally, democratising access to the best musculoskeletal care.

Rehumanizing Healthcare with AI: Beyond Efficiency to Effectiveness

A crucial aspect of Ashis’s vision is that digital transformation, particularly through AI, should not lead to “less humans” in healthcare. Instead, he believes it will allow healthcare professionals to “double down” on direct human interaction with patients, freeing them from burdensome backend processes that can be automated by AI and agents. This fundamental shift asks the “existential question: what needs to be human, what is best done by human, what is best done by automation?” 

Furthermore, Ashis stresses that AI’s potential extends beyond mere efficiency, which he acknowledges healthcare desperately needs. The second, often overlooked, ‘E’ in AI is effectiveness. He says – “There’s two E’s in AI and everybody forgets the second E. The first E is efficiency and everybody talks about efficiency. However, I think we miss the ball if we only focus on that one. And the second is effectiveness.“

He argues that healthcare must do better than it does today, addressing unmet needs, improving access, and ensuring people receive the “right care” more consistently. HSS’s commitment is not just to perform optimally but to codify that optimal approach and leverage technology to make healthcare more effective at delivering superior musculoskeletal care globally.

Movement: The Heart of Longevity and Healthcare

Finally, Ashis expands HSS’s broader vision beyond orthopaedics to movement itself, a cornerstone of quality of life. In a world focused on wearables and longevity, the ability to move freely is paramount. While much of healthcare focuses on loss associated with disease, musculoskeletal care represents gain—being able to play with grandchildren, run marathons, and live actively into old age. This vision aligns with Ashis’s hope for AI and digital transformation to “actually rehumanize healthcare” by preserving and enhancing the human capacity for life and movement.

Ashis Barad’s insights paint a compelling picture of a future where digital transformation, guided by clinical understanding and a clear vision for effectiveness, improves healthcare delivery fundamentally. His practical approach, rooted in personal experience and strategic focus, offers a roadmap for leveraging advanced technologies like Agentic AI to streamline operations, rehumanize patient experience, and democratize access to world-class care—impacting both the industry and lives worldwide.

From Lab to Life Sciences: Alicia’s Full-Circle AI Journey

Alicia’s work in AI began during her PhD at Columbia University in the 1990s, where she built a system to analyze radiographs and automatically generate radiology reports—a precursor to many of today’s AI diagnostic tools. That work led her to AT&T Bell Labs, where she contributed to cutting-edge speech and natural language processing research.

After 25 years at Bell Labs, she transitioned to Google Cloud, where she helped lead global AI strategy and worked on products like Vertex AI and Gemini. Her move to Novo Nordisk was driven by a desire to apply AI in a more purpose-driven setting. “I think pharma is still relatively new to AI, especially in commercialization. A little bit of governance, without being too heavy-handed, can help drive innovation and guide it in the right way for the right problems.”

Unlocking AI Use Cases in Pharma Commercialization

While most pharma AI initiatives have focused on drug discovery and clinical trials, Alicia sees immense opportunity in commercial operations—where AI can streamline processes, accelerate marketing execution, and improve personalization. Some of the most impactful AI use cases at Novo Nordisk include:

Knowledge Search and Retrieval: With massive volumes of data coming from internal reports, market intelligence, and vendor insights, finding relevant information can be overwhelming. Alicia’s team is implementing generative AI solutions with conversational interfaces that make it easy for marketing and insights professionals to extract answers—with references and source links.

AI-Powered Content Creation: Novo Nordisk is also using GenAI to automate ad copy variations, create short video content, and support early-stage creative development—dramatically reducing time-to-market for campaigns.

HCP Segmentation and Outreach: Traditional AI models help the company analyze prescribing behaviors, patient demographics, and treatment patterns—enabling smarter engagement strategies with healthcare providers.

These are real, tangible applications that deliver business impact—and Alicia is keenly focused on productizing only what adds real value.

Driving Cultural Change: The AI Ambassador Program

One of the biggest challenges in deploying AI at scale is change management. Alicia’s approach to this challenge is refreshingly people-first. Early in her tenure, she launched a “listening tour” to engage leaders across the organization and assess the current state of AI awareness and readiness.

She discovered a broad spectrum of sentiment—ranging from excitement to fear—driven by varying levels of understanding and concern around compliance. To address this, she developed the AI Ambassador Program, a grassroots initiative designed to foster peer-led learning and adoption. “In order to drive adoption of any product, you have to make it usable. You have to make the experience something that people will want to use—something intuitive, easy to use, that will drive adoption.”

The program has already exceeded expectations in participation. Ambassadors meet monthly to explore AI use cases, learn about emerging tools, and collaborate on safe experimentation within legal and compliance frameworks. It’s a model that blends top-down strategic intent with bottom-up enthusiasm—and one that Alicia believes can be replicated across industries.

Bringing a Product Mindset to AI Development

Another key differentiator in Alicia’s strategy is her commitment to a product management mindset to guide AI development at Novo Nordisk. Drawing from her experience at Google and AT&T, she’s brought a structured lifecycle approach to AI solution development: design, develop, test, monitor, iterate. Too often, she says, technologists get enamored with the possibilities of AI and rush to build solutions without fully understanding the problem. Her team starts with a foundational question: Should we be building this at all? Alicia adds, “Just because you can build something doesn’t mean you should.”

She emphasizes the need to prioritize AI initiatives only after validating user needs, business value, compliance implications, and technical feasibility does a solution move forward. This governance-light but insight-heavy approach ensures that Novo Nordisk focuses on building AI products that deliver real impact—not just innovation theater.

Partnering with Compliance from Day One

In an industry governed by strict regulations, compliance cannot be an afterthought. Alicia’s approach is to involve legal and data ethics teams early—during ideation, not after development. By building these partnerships proactively, she avoids project delays, ensures alignment with ethical standards, and creates a culture of responsible innovation.

“I told them, ‘You’re going to be my BFFs,’” she said. “I bring them in even when we’re just thinking about an idea—before any development starts. That way, we avoid surprises and ensure we’re building in a compliant manner.”

This early-stage partnership helps place guardrails that manage risk without stifling innovation—a delicate balance that’s essential in highly regulated industries.

Future Trends: LLMs, UX, and Human-AI Symbiosis

When asked about the future of AI, Alicia points to two major areas: user experience and contextual intelligence. “I think a trend that we’re going to see going forward… one of these big tech giants that are now in the business of creating large language models will now have that focus on the user experience.”

Alicia believes the success of ChatGPT wasn’t just about the model’s power—but the simplicity of the interface. As pharma builds its own AI tools, UX must remain a top priority to ensure adoption. But usability alone isn’t enough. She also anticipates a new evolution in AI models and adds that, “It’ll be interesting for how we see these large language models evolving—so that they go beyond just maybe text, image, and video, and start to bring in more contextual knowledge.”

She imagines a future where AI systems can incorporate real-world context and nuance like human collaborators. And the human-AI relationship is something Alicia reflects on deeply and states, “I think we still have that desire to see how do we make these machines behave maybe more like humans—without taking us out of the loop.”

This human-in-the-loop model is especially important in healthcare and pharma, where empathy, nuance, and ethical judgment matter.

A Playbook for Responsible AI in Pharma

Alicia Abella’s work at Novo Nordisk offers an inspiring model for how pharma companies—and other regulated enterprises—can responsibly scale AI. Her leadership showcases the importance of:

• A product-driven, outcome-focused strategy
• Strong compliance and legal collaboration
• Cultural change through education and empowerment
• A relentless focus on usability and trust

As more pharma companies explore AI applications beyond R&D, Alicia’s playbook provides a real-world guide for building AI programs that are credible, compliant, and customer-centric.

Rethinking Consumer Experience: Frictionless, Personalized, Proactive

Providence’s digital transformation efforts are centered on improving both patient and caregiver experiences by eliminating friction, enhancing personalization, and enabling proactive engagement.

“We really need to make finding our services and then transacting with us easier,” Sara noted. Whether it’s booking appointments, accessing financial counseling, or simply creating an account, patients should face minimal barriers. One of Providence’s priorities is enabling online scheduling for any bookable service—an effort aimed at meeting modern consumer expectations.

At the core of this strategy is personalization. Providence serves over five million patients annually across seven states, each with diverse needs and expectations. Sara emphasized that a one-size-fits-all approach no longer works. “Each person is different,” she said. “We need to recognize how important it is to speak to people in a way that keeps them engaged.”

Message Deflection Through Conversational AI

A key initiative that Sara highlighted is Providence’s use of conversational AI to deflect incoming messages that would otherwise burden physicians’ inboxes. The health system receives 6–7 million patient-generated messages annually, most of which are routed to physicians’ in-baskets.

Instead of optimizing around managing these messages, Sara and her team took a step back to ask: Why are patients sending these messages in the first place? What they found was that patients often couldn’t find the information they needed or complete basic tasks like booking appointments or understanding bills. Providence built a conversational and navigation platform to help patients resolve these issues in real time—without involving a physician.

This upstream solution has resulted in a 30% deflection rate, and Providence aims to deflect 2 million messages annually in the next few years. “It helps our patients get their needs met immediately, as opposed to having to wait 24 to 48 hours for a response,” said Sara. The success of this approach lies in combining AI agents with thoughtful product development, not simply layering on features.

Building a Digital Workforce: It’s Not Just About Automation

Sara cautioned against narrow interpretations of AI as merely a substitute for human labor. “With AI agents, it’s not just about automating the dull work,” she said. “It’s about doing things better.” Rather than automating low-value tasks for the sake of efficiency, Providence focuses on redesigning entire workflows for higher impact. “We don’t want to just automate crap,” Vaezy added bluntly. “We want to rethink processes from the ground up.”

This includes developing agents that can do things humans cannot—like analyzing massive datasets to identify the right individuals for targeted care outreach. “No human can parse through 10 or 20 million individuals to find the 1,000 people who need specific care,” she said. “AI can do that, and that’s where we see real value.”

Startup Incubation: DexCare and Praia Health

Providence has not just adopted new technologies; it has built them. Sara shared how the organization incubated and spun off two companies to address gaps in healthcare infrastructure.

DexCare, launched in 2021, focuses on supply-demand matching for on-demand care—ensuring patients find the right care, at the right time, in the right place. It helps patients discover appropriate services while balancing capacity on the provider side.

In 2024, Praia Health was launched to drive engagement through personalization. The platform helps deliver individualized digital experiences based on patient needs, preferences, and behaviors, rather than offering generic interactions.

These startups, both backed by venture capital, now operate independently while continuing to power Providence’s consumer-facing services. Vaezy credited earlier investments and organizational foresight during more financially stable times for enabling these ventures.

Designing for the Future: AI-Native Thinking

Looking ahead, Sara believes the next wave of innovation will require deeper integration of AI into business processes—moving from substitution to reinvention. “We’ll start to see more focus on what it looks like when something becomes AI-native, versus just being a tech overlay,” she predicted. For example, while ambient listening tools in clinical settings are generating excitement, Sara emphasized the importance of rethinking entire workflows to support these tools, rather than simply inserting them into outdated systems.

Another emerging area is observability—how organizations track, validate, and monitor AI systems in real-time to ensure safety and performance. “If you’re trying to run an unsupervised, non-deterministic model, you better have systems in place to make sure it’s not going rogue,” she said.

Responsible AI and Industry Accountability

Despite the excitement around generative AI, Sara urged caution and accountability. With increasing autonomy and the rise of AI agents, she emphasized the need for human-centered governance frameworks.

“It’s almost like fighting gravity,” she said. “Our job is to make this transition as responsible, humane, and ethical as possible.” She drew parallels to how consumers once expected to own their data—an ideal that faded in the face of corporate data control. “Let’s not miss the mark again,” she said. “We know this is going to happen. Now we have to ask: How do we do it right?”

Final Thoughts

Sara Vaezy’s insights offer a playbook for health systems navigating the shift to AI-native care delivery. Providence’s approach—centered on human needs, supported by intelligent systems, and grounded in ethical foresight—offers a compelling model for transformation.

By deflecting physician inbox overload, incubating purpose-built startups, and redesigning workflows with digital agents, Providence is not just implementing AI—it is rearchitecting healthcare delivery for the future.

As Sara said, “It’s not about reducing messages. It’s about the full experience.” In a rapidly evolving healthcare landscape, that mindset may be the key to unlocking AI’s true potential.

From “Meaningful Use” to Meaningful Impact

Aneesh Chopra reflects on the origins of the “meaningful use” program, which was designed to drive adoption of electronic health records (EHRs) through the HITECH Act. While the program succeeded in digitizing healthcare, Mr. Chopra acknowledges it fell short of transforming care delivery. Much of the technology was optimized for fee-for-service billing workflows rather than for improving clinical outcomes.

Now, Mr. Chopra sees a chance to realign health IT with the goals of value-based care. He emphasizes the need to define “meaningful use” in terms of patient outcomes, longitudinal care, and proactive health management. The renewed push, especially through the latest CMS-ONC RFI, presents an opportunity to finish what was started over a decade ago but with a clearer vision for outcomes-focused, AI-supported workflows. He urges the healthcare community to participate in shaping the future of health IT through these requests for information. “We want the technology to operate in a way that supports value-based care networks, helps consumers access their information, and enables real-time clinical decision-making,” he explains. Mr. Chopra emphasizes the need to move beyond back-office digitization to technology that enables smarter care delivery.

Introducing the “Healthcare Information Fiduciary”

One of the most innovative concepts Mr. Chopra introduces in the podcast is the idea of a “healthcare information fiduciary.” Drawing inspiration from financial fiduciary rules, this model proposes that applications and platforms handling patient data must act solely in the patient’s best interest.

In practice, this means creating a trusted marketplace of AI-enabled apps that help patients aggregate their medical records and receive personalized, unbiased recommendations. Such platforms would operate independently of the financial incentives of payers, providers, or pharmaceutical companies. “If you trust me with my information in a complex domain,” Mr. Chopra explains, “you must act in my best interest, and not based on how you get paid.” He states that this idea is already gaining traction through a voluntary code of conduct for consumer health apps, developed to encourage transparency around data use, including for de-identified data, an area traditionally excluded from HIPAA disclosures. This move toward greater openness and accountability is a step in the direction of building a full-fledged fiduciary model for healthcare data.

AI and Intelligent Workflows: Real-World Success Stories

Mr. Chopra shares compelling examples of how AI-driven workflows are already delivering tangible results. In one case, an academic medical center improved its Medicare Advantage star ratings by using conversational AI to reach out to patients, close care gaps, and ensure adherence to preventive care protocols. The campaign led to a significant quality improvement and unlocked over $6 million in incentive payments. These AI-powered agents contacted patients on behalf of their physicians, reminding them to complete necessary screenings or check-ups. By automating outreach at scale, the organization could more effectively engage patients—many of whom might otherwise fall through the cracks.

In another example, a health system leveraged AI-powered decision support tools to align with evidence-based guidelines. These tools acted as co-pilots for clinicians, reducing cognitive load and ensuring that treatment plans adhered to best practices. Though still in the pilot stage, this approach shows promise in enhancing care consistency and quality. The goal is not to replace clinical judgment but to support it with relevant, real-time data. AI tools can identify patients who match specific evidence-based criteria, flag care gaps, and help clinicians act quickly to address them. This type of augmentation could prove critical in improving care outcomes while reducing provider burnout.

Real-Time Data Through FHIR APIs

Interoperability has long been a challenge in healthcare, but Mr. Chopra is optimistic about recent progress. He points to the success of CMS’s FHIR API implementation, which is enabling near real-time access to claims data in programs like ACO REACH. This has transformed the utility of administrative data from retrospective analysis to proactive care management.

For example, advanced primary care providers working with high-risk populations are now able to detect changes in patient status within days of a clinical encounter, allowing them to act more swiftly. This real-time feedback loop represents a critical step toward building a learning health system that is both responsive and adaptive. Mr. Chopra explains that this shift reduces the lag between when a healthcare event occurs and when it can be addressed by a care team. Instead of waiting weeks or months for claims data to be analyzed, providers can now access that information in closer to real time, enabling more immediate interventions and better patient outcomes.

The Role of Public-Private Collaboration

Throughout the episode, Mr. Chopra emphasizes the importance of collaboration between the public and private sectors. His own career reflects a passion for creating handshakes and handoffs where policy creates the framework, and the private sector drives innovation.

He cites the CARIN Alliance, a bipartisan initiative he co-founded, as an example of progress in creating a voluntary code of conduct for consumer health apps. This code aims to increase transparency around data use, including de-identified data, and build consumer trust. According to Chopra, the combination of data sharing rights and an AI code of conduct will catalyze a new generation of responsible, patient-centered tools.

According to Mr. Chopra, the combination of data sharing rights and an AI code of conduct will catalyze a new generation of responsible, patient-centered tools. He also credits the efforts of bipartisan policymakers and industry leaders in sustaining progress over time. “We’ve seen administrations change, but the momentum toward smarter health data use continues,” he says. This consistency is essential for driving lasting innovation and achieving meaningful outcomes at scale.

From Vision to Action: Enabling the Future of Healthcare

Mr. Chopra closes the conversation with a rallying cry to the healthcare innovation community. He urges early adopters, across startups, health systems, and technology vendors, to raise their hands and help test, validate, and scale the tools needed to support a more equitable and efficient healthcare system. “If you see the future and you want to have a hand in bringing it to life,” he says, “we would love to tap into that talent.”

He emphasizes that meaningful transformation will require experimentation, feedback, and iteration. The public sector is opening the door with policies and programs but it is up to the private sector to walk through it and deliver results. Early adopters will play a vital role in shaping the next chapter of healthcare innovation.

This episode of The Big Unlock is more than a retrospective on health IT policy, it is a forward-looking manifesto for how data, AI, and innovation can reshape American healthcare. Aneesh Chopra’s insights serve as a roadmap for leaders seeking to bridge the gap between technological capability and real-world outcomes. Whether you’re a policymaker, provider, technologist, or entrepreneur, there’s a clear message – the future of healthcare lies in building trusted, intelligent, and patient-first systems and the time to act is now. By learning from past efforts and applying the best of today’s technologies, healthcare stakeholders have an opportunity to co-create a future where high-quality care is equitable, accessible, and guided by data we can trust.

From Ophthalmology to AI Leadership

Dr. Liu’s foray into AI began in the late 2010s during his clinical training, sparked by groundbreaking studies—particularly one from Google—that demonstrated the ability of AI models to predict cardiovascular risk factors from retinal images. This superhuman diagnostic capability was a turning point for him. As a retina specialist immersed in an image-rich field, Dr. Liu recognized the untapped potential of deep learning to transform how clinicians interpret complex visual data.

At Johns Hopkins, Dr. Liu leads the Gills AI Center—the first endowed AI initiative at the Johns Hopkins School of Medicine—while also maintaining an active clinical practice. He contributes across four pillars: AI development, implementation, governance, and scientific innovation, giving him a panoramic view of the opportunities and challenges in healthcare AI.

Autonomous AI in Primary Care: A Case Study in Diabetic Retinopathy Screening

One of the most compelling examples Dr. Liu shared was the deployment of an FDA-approved autonomous AI system to detect diabetic retinopathy in primary care settings. This system was the first of its kind to be approved for autonomous clinical use, and Johns Hopkins began implementing it in 2020.

Traditionally, patients needed to see a separate specialist to complete an annual retinal screening—an extra step that often led to missed appointments and lower screening rates. The AI system allows primary care physicians to take retinal images in their office, with AI analyzing them in real time. Patients receive immediate results, and only those with positive screenings are referred to an ophthalmologist.

The outcomes have been striking. Johns Hopkins observed a marked improvement in screening adherence, especially among underserved populations such as African Americans and Medicaid recipients. These results, published in Nature Digital Medicine, underscore how AI can help close gaps in preventive care—if implemented thoughtfully.

Generative AI for Revenue Cycle: From Clinical to Operational AI

AI’s impact at Johns Hopkins isn’t limited to the clinic. Dr. Liu described a pilot project using generative AI for revenue cycle management, specifically prior authorization. This is a high-friction area in healthcare, involving extensive paperwork and delays in care.

By leveraging large language models (LLMs), Johns Hopkins automated prior authorization workflows, reducing the time required and handling unstructured data far more effectively than traditional robotic process automation (RPA) methods. These results illustrate how AI can unlock value beyond clinical domains by streamlining healthcare operations and improving provider efficiency.

Startups and the Reality of Healthcare AI

Drawing from his experience working with numerous startups, Dr. Liu offered candid advice to AI entrepreneurs: understand reimbursement from day one. “I think one of the common mistakes that startup companies make in the healthcare AI space is not considering or not understanding their reimbursement issue from day one,” Dr. Liu added. Many startups make the mistake of focusing on building a great product without planning for how it will be paid for—especially in a field as complex and regulated as healthcare. 

He emphasized that FDA approval alone isn’t enough. Startups must also determine whether existing CPT codes apply to their solution, and if not, navigate the lengthy and uncertain process of obtaining new ones. Beyond regulatory hurdles, they must build business models that reflect the real-world economics of health systems.

Startups often underestimate the cost of this journey—$3 to $5 million for FDA approval is typical—and many don’t budget appropriately. Dr. Liu’s message was clear: clinical AI solutions need sound financial strategies as much as innovative technology.

Creating Enterprise-Ready AI: The Johns Hopkins Governance Model

To manage the influx of AI tools and ensure responsible adoption, Johns Hopkins established a robust AI governance framework. Dr. Liu is part of an eight-member enterprise leadership team that evaluates all AI-related initiatives across the health system.

This governance model is built around seven core principles: fairness, transparency, accountability, ethical data use, safety, evidence-based effectiveness, and sustainability. Any AI vendor seeking to partner with Johns Hopkins must complete a standardized intake process, provide detailed documentation on their tool’s safety, ROI, and evidence base, and undergo a rigorous review process.

The system categorizes tools based on their use case—clinical, operational, or imaging—and advances each proposal through specialized review committees. This ensures that tools align with Johns Hopkins’ mission, technical infrastructure, and patient care goals before they are deployed at scale.

This governance model could serve as a blueprint for other integrated health systems navigating a crowded and often chaotic AI vendor landscape.

Looking Ahead: Omics, Risk Prediction, and Scaling Innovation

Dr. Liu also shared his excitement about the emerging field of AI-driven “omics,” particularly using retinal biomarkers to predict systemic health conditions such as cardiovascular disease, kidney damage, and dementia. AI-enabled retinal screening programs in community settings could identify at-risk individuals years before symptoms emerge.

However, he was quick to point out that identifying risk is only part of the equation. Health systems must also build the care pathways to ensure those flagged by AI are connected to the appropriate subspecialists and receive timely follow-up care. Without that, the potential of predictive AI will remain unrealized.

A Call for Collaboration: Startups, VCs, and Health Systems

In his closing remarks, Dr. Liu highlighted a growing but still insufficient level of collaboration between AI startups, venture capitalists, and integrated health systems. Startups drive innovation and speed—but they often lack the domain knowledge and infrastructure to scale safely. Health systems, on the other hand, deliver the majority of care but tend to move slowly due to regulatory and operational constraints.

Bridging this gap, he argued, is essential for sustainable AI deployment. Startups need to understand the realities of clinical practice and reimbursement. Health systems need to improve agility and decision-making. And investors need to align their expectations with the long, complex arc of healthcare innovation.

Dr. Liu hopes to see more structured partnerships where these groups work together to solve real problems, share risk, and scale proven solutions responsibly. He believes that such collaboration is essential for delivering long-term value—and ultimately, for improving health outcomes.

AI is Here to Stay

As Dr. Liu puts it, “The train has left the station.” AI is already reshaping healthcare, and the focus must now shift to responsible scaling, thoughtful implementation, and real-world results. “ I think the vast majority of people will agree that AI will change medicine and society as we know it,” he adds. 

Whether through autonomous diagnostic tools, generative AI for operational efficiency, or predictive omics models, the future of healthcare will be defined by our ability to integrate AI into the fabric of care—ethically, equitably, and effectively.

This episode is a powerful reminder of what it really takes to turn promising AI into real-world results. For health systems, startups, and investors, Dr. Liu’s insights highlight why successful innovation depends as much on execution as on technology.