Why “Disruptive” AI Won’t Work in Healthcare
Disruption vs. Strategic Innovation in Guiding Healthcare AI Applications
Healthcare and health tech industries have been abuzz with talk of “disruption” and the transformative potential of artificial intelligence (AI), particularly generative AI. While these technologies undoubtedly hold immense promise, the rhetoric surrounding their implementation often oversimplifies the complex realities of healthcare systems.
This article explores the nuanced landscape of healthcare innovation, arguing for a more measured, strategic approach to integrating AI and other advanced technologies into medical practice.
The Allure of Disruption
The term “disruption” has become a buzzword in tech circles, often used to describe innovations that radically alter existing markets or create entirely new ones. In healthcare, proponents of disruptive technologies promise revolutionary changes that will dramatically improve efficiency, reduce costs, and enhance patient outcomes. Generative AI, in particular, has been hailed as a potential game-changer. Its ability to analyze vast amounts of unstructured data, generate human-like text, and assist in complex decision-making processes has captured the imagination of healthcare leaders and investors alike.
The excitement is understandable, given the potential applications across various aspects of healthcare, from administrative tasks to clinical decision support and drug discovery. However, the enthusiasm for disruption often overlooks the unique challenges and responsibilities inherent in healthcare delivery. Unlike many other industries, healthcare deals directly with human lives and well-being, operating under strict regulatory frameworks and ethical guidelines. The cost of failure or unintended consequences in this field can be catastrophic.
The Kuhnian Perspective on Scientific Progress
To better understand the nature of change in healthcare, it’s useful to consider Thomas Kuhn’s concept of scientific revolutions. Kuhn, a philosopher of science, in 1962 argued that scientific progress occurs through periods of “normal science” punctuated by paradigm shifts or “revolutions.” While Kuhn used the term “revolutions,” his description of scientific progress aligns more closely with what we might call strategic transformations:
- Normal Science: Most scientific work occurs within established paradigms, making incremental progress.
- Puzzle-Solving: Scientists work within the established framework to solve specific problems.
- Accumulation of Anomalies: Scientific revolutions often result from the gradual accumulation of anomalies that the current paradigm struggles to explain.
- Paradigm Shift: The actual “revolution” is more of a shift to a new paradigm that better explains observed phenomena rather than a sudden, complete overhaul.
Kuhn’s perspective suggests that even significant changes in scientific understanding are often the result of gradual accumulation of evidence and incremental shifts in thinking rather than sudden, disruptive breaks with the past. This view aligns well with the complex nature of healthcare systems, where change often requires careful maneuvering within existing structures.
Fitting AI Into the Puzzle of Healthcare
You may have a memory of a 15-tile puzzle game from when you were a kid or similar digital versions that exist today. This classic sliding puzzle is an apt metaphor for the complexity of healthcare innovation, particularly when implementing artificial intelligence (AI) and other advanced technologies.
In healthcare, like in this sliding puzzle game, leaders face a fixed frame with limited space to maneuver. Healthcare organizations often operate within tight constraints of resources, regulations, and established workflows, restricting their ability to implement sweeping changes. Each “move” in the innovation process must be carefully considered, as every decision affects the entire system. Just as solving a sliding puzzle requires a methodical approach, healthcare innovation demands a strategic, step-by-step implementation rather than abrupt, disruptive changes.
The interconnected nature of healthcare systems adds another layer of complexity to this puzzle. Implementing new AI technologies impacts various aspects of the organization, from workflows to staff roles and patient care. Leaders must maintain a clear vision of the end goal while remaining adaptable to evolving challenges. This multidimensional approach requires engaging across technological, strategic, operational, and organizational domains, much like considering both horizontal and vertical movements in a sliding puzzle. Ultimately, solving this healthcare innovation puzzle demands patience and persistence, with realistic timelines often spanning 3–5 years due to the complex interplay of stakeholders and systems.
The Case for Strategic Innovation in Healthcare
Drawing on Kuhn’s insights, we argue that healthcare innovation, particularly in the realm of AI application development and integration, should be approached as a process of Strategic Transformation rather than disruptive upheaval. This approach recognizes the complexity of healthcare systems and the need for careful, deliberate, and ethical integration of new technologies.
Kuhn’s framework highlights that transformation in complex systems, like healthcare, is rarely a rapid, all-encompassing disruption. Instead, meaningful progress often unfolds as incremental advancements within an existing paradigm, punctuated by occasional shifts when innovations prove their utility and alignment with systemic needs. This nuanced understanding of change aligns well with the challenges of implementing AI in healthcare, where each innovation must be carefully integrated into the existing “puzzle” of healthcare delivery.
Using the sliding puzzle game analogy, technological disruption in healthcare would be akin to pulling all the puzzle pieces out and shoving them back into the frame to get the “final solution” rather than incrementally sliding them into place with the correct order.
Bridging Theory and Practice: A Framework for Healthcare AI Integration
The potential for disruption is not uniform across technologies. Instead, it varies based on the complexity of the technology, the infrastructure it requires, and the level of change it introduces to existing workflows. To provide a practical perspective, we propose a 2D matrix that categorizes healthcare AI technologies based on their degree of disruption and the complexity of healthcare infrastructure and technology needs.
This matrix can serve as a strategic guide for evaluating which technologies are ready for integration, which require more foundational groundwork, and how organizations can prioritize investments to balance innovation with sustainability.
Healthcare AI Technology, Complexity and Potential for Disruption
Here, we explore a variety of envisioned ML/AI applications and integration initiatives in healthcare. As part of the strategy for planning potential innovation projects, we propose an assessment process that is use-case and sensitive to the dynamics of healthcare systems.
Figure 1 highlights the varying demands placed on healthcare organizations based on the degree of likely technological disruption and infrastructure complexity from each listed AI technology integration. This scatter plot provides a two-dimensional representation of healthcare AI technologies, focusing on the relationship between their degree of disruption and the complexity of healthcare infrastructure and technology needs required for their deployment.
This matrix can be viewed as a representation of the sliding puzzle of healthcare innovation, where each technology represents a piece that must be carefully maneuvered into place within the existing healthcare landscape. Implementing these technologies should be approached strategically, considering their potential impact and the resources required for successful integration.
Low Potential for Disruption, Lower Complexity
RPA (Robotic Process Automation) has low disruptive potential as it typically automates routine, rule-based tasks without significantly altering existing workflows. It’s relatively simple to implement, often integrating well with existing systems and tech infrastructure.
Basic ML/Analytics tools have low disruptive potential as they often enhance existing processes rather than replace them. They’re generally straightforward to implement, leveraging existing data infrastructure.
Medium Potential for Disruption, Moderate Complexity
Robust AI Governance: While not directly disruptive to clinical practices, implementing robust AI governance can significantly impact organizational processes and decision-making. It requires careful planning and integration across various departments.
Predictive AI has moderate disruptive potential as it can significantly enhance decision-making processes. Implementation complexity is moderate, requiring integration with existing systems and workflows.
Basic AI integrations with Electronic Health Records (EHRs) have moderate disruptive potential, enhancing existing systems rather than completely overhauling them. Implementation complexity is moderate, requiring careful integration with existing EHR systems.
RAG (Retrieval-Augmented Generation) systems have moderate disruptive potential, enhancing information retrieval and generation capabilities. Implementation complexity is moderate, requiring integration with existing knowledge bases and systems.
Intermediate AI solutions, falling between basic and advanced implementations, have moderate disruptive potential and complexity, depending on the specific application.
Cybersecurity Enhancements for AI. While not directly disruptive to clinical practices, implementing AI-enhanced cybersecurity can significantly impact organizational processes. It requires careful integration with existing security infrastructure.
IoT Integrations have moderate disruptive potential, introducing new data sources and monitoring capabilities. Implementation complexity is moderate, requiring integration with existing systems and new device management.
Many healthcare organizations, like hospitals or ambulatory care centers, already use predictive AI for specific functions and have IoT integrations for some devices. Enterprise-wide integrations and expansion of predictive AI with advanced generative AI outputs, like triage systems, will push the complexity assessment and disruption potential further up the scales.
High Potential for Disruption, High Complexity
GenAI Applications have high disruptive potential, transforming various aspects of healthcare from diagnosis to treatment planning. Implementation complexity is high, requiring significant infrastructure and workflow changes.
Agentic AI (Multi-agent) has high disruptive potential, potentially revolutionizing complex decision-making processes in healthcare. Implementation complexity is high, requiring sophisticated infrastructure and integration.
Advanced AI integrations with EHRs have high disruptive potential, potentially transforming how healthcare providers interact with patient data and make clinical decisions. Implementation complexity is high, requiring significant changes to existing EHR systems and workflows.
While these technologies may appear to be fairly straightforward in the pilot stage, getting them to scale requires significant infrastructure investments and will have an increased disruptive potential with enterprise-wide implementation.
Deployment Models and Healthcare Contextual Factors
Another consideration is the deployment models that would ideally be implemented when integrating these specific technologies into a healthcare system. Most healthcare organizations already operate with a mix of on-premise, cloud-based, and hybrid tech stacks and data solutions. New AI deployments and integrations will necessitate reevaluating this mix to determine the optimal balance of resources required.
It is important to note that this visualization is based on assumptions about how healthcare organizations, on average, might deploy these technological solutions and integrations. For larger and more complex healthcare organizations — such as those with multiple campuses or geographically dispersed sites — the distribution of technologies across this matrix may shift. For example:
- Infrastructure and Deployment Variability: Larger organizations may lean more heavily on hybrid models to accommodate diverse IT environments and ensure compliance with local data regulations.
- Adoption of High-Disruption Technologies: These organizations might prioritize investments in cloud-based solutions for scalability while also maintaining on-premise components for critical systems, databases, and backups.
These considerations further illustrate the complexity of the healthcare innovation puzzle, where each piece must fit within the organization and broader patient population mix, staffing resources, and regulatory and technological landscapes.
Gradual Paradigm Shifts vs. Disruption
The concept of gradual paradigm shifts is particularly relevant to healthcare, where abrupt changes can pose significant risks. The healthcare sector, with its complex regulatory environment, established practices, and critical focus on patient safety, is not conducive to sudden, disruptive changes. Also, for good reason, healthcare providers are often resistant to changes that involve adopting new technologies or externally-developed systems.
For generative AI to be successfully integrated into healthcare, we should approach it as a strategic innovation rather than a disruptive force. This approach involves:
- Integration with Existing Systems: AI applications must work seamlessly with current electronic health records (EHRs) and other healthcare technologies and workflows.
- Ethics and Patient-Centricity: AI tool development must have an ethical foundation and guardrails, and their implementation must adhere to regulatory guidelines and prioritize patient needs, privacy, and safety.
- Incremental Deployment: A phased approach allows for careful model evaluation, iteration, and adjustment of data inputs at each stage across the AI lifecycle.
Not surprisingly, this strategic approach also aligns with the sliding puzzle metaphor, where each innovation must be carefully positioned within the existing healthcare framework.
The Case for Strategic AI Innovations in Healthcare
Adopting a strategic transformation approach to AI implementation offers a nuanced and sustainable path forward. Rather than chasing the latest AI technology for its own sake, this approach focuses on aligning AI initiatives with organizational needs, operating environment, and capabilities.
Key Advantages of the Strategic Transformation Approach
- Mitigated Risks: Incremental improvements and iterative learning reduce the potential for costly AI missteps.
- Greater Agility: Building a culture of adaptability positions organizations to navigate evolving market and AI regulatory landscapes.
- Stronger Stakeholder Alignment: Clear articulation of AI’s business value fosters buy-in from executives, staff, and patients.
- Sustainable Progress: Strategic innovation ensures long-term growth, avoiding the pitfalls of short-term, reactive implementations.
By emphasizing deliberate, incremental advancements over disruptive upheaval, healthcare organizations can responsibly harness AI's transformative potential. In turn, healthcare organizations can drive meaningful, lasting change while safeguarding the well-being of patients and the integrity of the system.
Timeframe for Implementation
Realistic expectations for significant changes in healthcare often span 3–5 years because of the complex interplay of provider, institutional, payor, regulator, and patient-driven imperatives. While there’s excitement about GenAI’s potential in healthcare, there’s also a necessary caution. We’ve previously written about how AI strategy mapping and use-case prioritization are essential first steps. Developing AI strategy and evaluation across the entire lifecycle allows for robust governance frameworks while also ensuring flexibility for:
Extensive Testing and Validation: Ensuring the reliability and safety of AI systems in clinical settings and with context-driven and evolving workflows.
Stakeholder Adaptation: Giving healthcare professionals time to train, adapt, and adjust to new AI-assisted workflows.
Regulatory Alignment: Allowing regulatory frameworks to evolve alongside technological advancements.
Healthcare Strategic Innovations for 2025- “Predictions”
The strategic innovations driving AI evolution in 2025 focus on building a solid evidence base for practical, industry-specific applications. This is particularly crucial in regulated industries like healthcare, finance, manufacturing, and beyond. The “trends” likely to dominate in 2025 include:
Localized AI Models: Advancements in small language models (SLMs) and multimodal large language models (MMLMs) tailored to specific industries. These models will integrate local data and context to enhance relevance and accuracy, particularly in compliance-heavy sectors like healthcare and finance.
Agentic AI and Real-Time Adaptation: Integrating retrieval-augmented generation (RAG) and agentic frameworks to allow AI systems to pull from real-time data sources. This ensures recommendations and insights remain up-to-date and contextually appropriate, a critical requirement for most healthcare applications.
Human-in-the-Loop (HITL) Optimization: Evolving HITL processes to enable adaptive feedback loops where AI-generated outputs are refined based on domain expert input. This makes AI more reliable and actionable in high-stakes decision-making, from clinical trials to crisis response.
Cross-Industry Benchmarks: Establishing standardized frameworks to test AI systems across multiple domains. For instance, MMLMs can be tested in both healthcare diagnostics and legal document analysis to uncover shared capabilities and identify domain-specific adjustments.
Ethics-Driven Innovation: Creating tools and methodologies for AI risk assessment and governance to ensure compliance, safety, and societal acceptance. AI systems will increasingly embed ethical and regulatory guardrails, especially as they scale into public-facing roles.
These innovations represent new pieces in the healthcare sliding puzzle, each requiring careful integration into existing systems and workflows.
The Role of Leadership in Strategic Innovation
Leaders in healthcare AI are not just decision-makers but the architects of a new era in patient care. Imagine a hospital leadership navigating the complex sliding puzzle, where each piece represents a critical component of the healthcare system — from existing workflows to new AI technologies. Like in the 15-square puzzle, leaders must strategically shift these pieces within a confined space, carefully maneuvering each element to create a cohesive picture of enhanced patient care and operational efficiency.
This sliding puzzle metaphor aptly captures the challenges healthcare leaders face. With limited room for maneuvering and numerous interconnected pieces already in place, they must make calculated moves to integrate AI innovations without disrupting the entire system. Each shift requires foresight and precision, as a misplaced piece could throw the entire puzzle into disarray.
Leaders must articulate a clear vision for how AI will enhance patient outcomes and organizational performance, all while working within the constraints of existing structures. They are tasked with allocating resources judiciously, making high-stakes decisions on technology investments, data infrastructure upgrades, and comprehensive staff training programs. Like a puzzle solver, they must balance the immediate need to create space for new elements with the long-term goal of achieving a harmonious, complete picture.
By fostering a culture that embraces AI while maintaining an unwavering focus on compassionate, human-centered care, healthcare leaders can transform their organizations into models of strategic innovation.
The question remains: can today’s healthcare leaders solve this intricate puzzle, methodically sliding each piece into place to reveal a transformed, AI-enhanced healthcare landscape?
Conclusion: The Path Forward
Integrating AI into healthcare represents significant opportunities to enhance patient care, improve operational efficiency, and drive medical innovation. However, realizing this potential requires a strategic, measured approach that respects the complexities and responsibilities of healthcare delivery.
By moving beyond the rhetoric of technological disruption and embracing a model of strategic innovation, healthcare organizations can harness the power of AI while maintaining their commitment to patient safety, ethical practice, and high-quality care. This approach, grounded in careful planning, incremental implementation, and ongoing evaluation, offers a more sustainable and responsible path to healthcare transformation.
Drawing from Kuhn’s framework and the sliding puzzle metaphor, we recognize that meaningful progress in healthcare AI integration will likely occur through a series of gradual paradigm shifts rather than abrupt disruptions. This approach aligns with the complex nature of healthcare systems and the critical importance of maintaining patient safety and trust throughout the innovation process.
As we move forward, it’s essential to view AI implementation not as a race to disrupt but as a carefully orchestrated journey of strategic transformation that respects the unique challenges and responsibilities of the healthcare sector.
AI startups, developers, and healthcare organizations can minimize the disruptive potential of AI integrations through adequate AI strategic planning and a robust governance framework.
Find out how Health-Vision.AI, LLC can help guide you.
