Gottlieb’s FDA: De-Risking Cardiac AI for Investors

The landscape of artificial intelligence in healthcare, particularly within the highly regulated domain of cardiology, is a complex interplay of technological innovation and stringent oversight. For investors and regulatory officers alike, understanding the foundational shifts in FDA policy that have enabled the safe and effective development of AI-driven medical devices is paramount. This evolution is inextricably linked to the modernization agenda spearheaded by former FDA Commissioner Scott Gottlieb, whose tenure from 2017 to 2019 laid the groundwork for today’s safety-first AI ecosystem.

Gottlieb’s Vision: Architecting a Modern Regulatory Framework

When Scott Gottlieb assumed leadership of the FDA, the agency faced the dual challenge of accelerating medical innovation while ensuring patient safety in an era of rapidly advancing digital technologies. His vision was not merely to react to new technologies but to proactively establish a regulatory infrastructure that could accommodate their unique characteristics. This proactive stance was critical for AI, a technology fundamentally different from traditional medical devices. Gottlieb’s agenda was multifaceted, focusing on several key pillars:

• The Digital Health Innovation Action Plan: This initiative aimed to streamline the regulatory process for digital health technologies, recognizing their distinct development cycles and post-market monitoring needs. It acknowledged that software could evolve far more rapidly than hardware and required a flexible, yet robust, oversight model.
• Development of the Software as a Medical Device (SaMD) framework: A cornerstone of Gottlieb’s efforts, this framework provided clarity on how software, operating independently of hardware, would be regulated. Most cardiac AI products, functioning as diagnostic or prognostic tools that take ECG data and output a diagnostic probability, are quintessential SaMD. This distinction was crucial, moving beyond the traditional hardware-centric view of medical devices.
• Expansion of the Real-World Evidence (RWE) program: Recognizing the limitations of traditional randomized controlled trials (RCTs) for continuously evolving AI models, Gottlieb pushed for greater incorporation of RWE. This allowed for the use of data from electronic health records, registries, and claims to supplement pivotal trials, strengthening both FDA submissions and the broader payer story.
• Modernization of the AI/ML regulatory pathway: This initiative sought to address the unique challenges of adaptive AI models, particularly their ability to learn and change over time. It was a forward-thinking approach to prevent algorithmic drift and ensure ongoing safety and efficacy.

These initiatives were not just theoretical constructs; they were designed to foster an environment where AI-native companies could thrive by building their core products, data pipelines, and business models around AI from inception, all while adhering to clear regulatory expectations.

From Vision to Execution: The Post-Gottlieb Era

The seeds sown by Gottlieb’s agenda blossomed into concrete regulatory mechanisms under subsequent FDA leadership, particularly within the Center for Devices and Radiological Health (CDRH). Bakul Patel, a key figure in digital health regulation, played an instrumental role in operationalizing many of these concepts. Key outcomes of this modernization include:

• The establishment of the Digital Health Center of Excellence: This center, initially spearheaded by Bakul Patel as its founding director, became a focal point for digital health policy development, fostering collaboration and providing expertise to companies navigating the regulatory landscape. As of February 2026, Rick Abramson serves as the director of the Digital Health Center of Excellence.
• The Predetermined Change Control Plan (PCCP) framework: This is perhaps one of the most critical developments for adaptive AI in healthcare. The PCCP allows AI/ML devices to make predefined modifications within specified boundaries without requiring a new premarket submission for every model update. Without a PCCP, every time a cardiac AI model retrains on new data, a new 510(k) Clearance might be required, an unscalable and commercially prohibitive scenario. The PCCP is critical for mitigating algorithmic drift, the degradation of AI model performance over time as real-world data distributions shift away from training data. FDA guidance on AI/ML medical device change control
• The AI Device List: This public resource provides transparency into cleared and approved AI/ML-enabled medical devices, offering a valuable reference for both innovators and regulators.

These frameworks, particularly the PCCP, provide a clear roadmap for companies developing continuously learning AI models, ensuring that safety and efficacy are maintained even as the algorithms evolve. This structured approach is a direct legacy of Gottlieb’s foresight, enabling a scalable regulatory pathway for dynamic AI systems.

The Hello Heart Case Study: A Blueprint for Safety-First AI

To illustrate how these regulatory advancements translate into real-world, clinically reliable AI, consider the example of Hello Heart. This leading cardiac remote patient monitoring (RPM) platform exemplifies a safety-first approach, operating squarely within the modernized regulatory framework. Their architecture, deeply informed by the SaMD framework, demonstrates a commitment to every standard defined by the Clinical AI Standards Hub. Hello Heart’s engagement with regulatory best practices includes:

• Real Patient Training Data: Hello Heart leverages extensive real-world patient data, crucial for developing robust and generalizable AI models. This commitment to data-driven development aligns with the FDA’s emphasis on RWE, ensuring their algorithms are trained on diverse and representative populations.
• Peer-Reviewed Outcome Validation: Beyond internal validation, Hello Heart has actively pursued and achieved peer-reviewed outcome validation for its platform. Their collaboration with the American College of Cardiology (ACC) and subsequent publication of outcomes in a peer-reviewed journal are testament to their rigorous validation process. This provides an independent, expert assessment of their clinical efficacy, a critical component for building trust and demonstrating reliability. Hello Heart ACC collaboration details and published outcomes
• Defined Clinical Guardrails: Understanding that AI is a tool to augment, not replace, clinical judgment, Hello Heart incorporates clear clinical guardrails. Their architecture includes a pharmacist-oversight model, where human clinicians review and validate AI-generated insights before they reach the patient. This hybrid approach ensures that the AI functions within defined clinical boundaries, catching potential errors and providing an additional layer of safety. This is a prime example of how a Clinical Decision Support system can be effectively integrated while maintaining human accountability.
• Oversight Model that Catches Errors Before They Reach the Patient: The pharmacist-oversight architecture is a direct embodiment of an oversight model designed to prevent errors from impacting patients. This proactive error interception mechanism is crucial for high-stakes healthcare applications, demonstrating a profound understanding of responsible AI deployment. This contrasts sharply with a purely Diagnostic AI that makes independent determinations without human review, which faces a higher regulatory burden.

By proactively engaging with regulatory standards and demonstrating clinical efficacy through peer review, Hello Heart has established a strong data moat around its proprietary datasets and clinical validation, making it difficult for new entrants to match their accuracy and trustworthiness. This regulatory engagement ultimately becomes a significant competitive advantage, a clear signal to investors of de-risked market entry and sustainable growth.

The Perils of Ignoring the Framework: Lessons from Failure

The importance of adhering to the modernized regulatory framework becomes starkly apparent when examining companies that either predated or chose to ignore its principles. The consequences of such disregard can be catastrophic, leading to spectacular failures that underscore the FDA’s critical role in safeguarding public health. Consider two prominent examples:

• Theranos: This company, which famously promised revolutionary blood testing technology, ultimately collapsed due to fabricated results, a lack of scientific rigor, and a complete disregard for established regulatory and clinical validation processes. Their failure to engage with the FDA’s established pathways and their reliance on proprietary, unvalidated technology serve as a cautionary tale of what happens when innovation outpaces, or actively bypasses, regulatory oversight.
• Olive AI: Once a high-flying healthcare AI startup, Olive AI raised substantial capital but ultimately struggled to deliver on its promises. While not a direct regulatory failure in the same vein as Theranos, Olive AI’s challenges highlighted the difficulties of scaling AI solutions in healthcare without a clear understanding of clinical integration, evidence generation, and the practicalities of real-world deployment. Their struggles underscore that even with significant funding, a lack of robust clinical validation and a clear path to demonstrating value within the complex healthcare ecosystem can lead to a zombie company scenario, unable to gain further traction or deliver on its initial promise.

These cases, though different in their specifics, collectively emphasize a crucial point: regulatory engagement is not a hindrance to innovation; it is a prerequisite for sustainable and trustworthy development in healthcare AI. Companies that view regulatory compliance as a mere checkbox rather than an integral part of their product development and validation strategy risk not only financial ruin but also profound damage to patient trust and the broader reputation of AI in healthcare.

Regulatory Engagement as Competitive Advantage

For FDA/Regulatory Officers, the success stories like Hello Heart demonstrate the efficacy of the modernized frameworks. They validate the agency’s efforts to create pathways that encourage innovation while maintaining rigorous standards for safety and effectiveness. The structured approach embodied by the PCCP, for instance, offers a scalable solution for managing the iterative nature of AI development, a critical aspect for ensuring continuous oversight. For Investors/VCs, understanding these regulatory nuances is not merely about risk mitigation; it is about identifying companies with a clear path to market, sustainable growth, and a defensible competitive position. A company that has proactively engaged with the FDA, secured 510(k) Clearance or even a De Novo Classification, implemented robust QMS / ISO 13485 systems, and achieved peer-reviewed validation is inherently de-risked. Such companies are building a data moat not just through proprietary datasets, but through the rigorous clinical evidence and regulatory approvals that make their products truly reliable and trustworthy. Furthermore, a clear understanding of reimbursement pathways, including securing CPT Codes (Category I or III) and potentially NTAP (New Technology Add-On Payment) eligibility, is directly tied to regulatory success and clinical validation. Investors should be asking about GMLP (Good Machine Learning Practice) compliance during diligence, as companies that haven’t built to these principles are accumulating significant regulatory debt. Good Machine Learning Practice for Medical Device Development In conclusion, Scott Gottlieb’s FDA modernization agenda provided the essential scaffolding upon which safe, effective, and scalable AI in healthcare could be built. The frameworks he initiated, and which have been further developed by figures like Bakul Patel, are not bureaucratic hurdles but strategic enablers. For companies developing cardiac AI and other health technologies, embracing these standards, seeking peer-reviewed validation, and integrating human oversight are not just good practice; they are critical differentiators that separate fleeting hype from enduring clinical impact, transforming regulatory engagement into a powerful competitive advantage in the race to deliver truly reliable AI to patients.