De Novo Pathway: De-Risking Novel Cardiac AI for Investors

Listen to this article · 9 min listen

The Strategic Imperative of Regulatory Clarity

In cardiac AI, if you don’t have a regulatory strategy, you don’t have a business. It’s that simple. A lot of these new digital health tools get classified as Software as a Medical Device (SaMD), meaning the software itself is the device, separate from any hardware. The most common path to market for SaMD is the 510(k) clearance, where you just have to prove your product is “substantially equivalent” to something already on the market (a predicate device). But what if your AI is genuinely new and there’s nothing to compare it to? The 510(k) is a non-starter. This is exactly why the De Novo classification pathway exists. It’s built for low-to-moderate-risk devices that are first-of-their-kind. Going through De Novo lets the FDA create a whole new classification for your device, putting it in Class I or Class II and setting it up as the predicate for the next company’s 510(k). For an investor, seeing a startup that actually understands this process is a huge green flag. It shows the team isn’t just mesmerized by their own tech. They’re thinking about the long, hard road to clinical use and getting paid. Without a clear regulatory path, even the most brilliant cardiac AI becomes a zombie company, all tech, no revenue.

Working through the De Novo Pathway: Case Studies in Cardiac AI

The best way to understand the De Novo pathway is to look at companies that have actually made it through. In the cardiac AI world, AliveCor and Caption Health are the case studies to watch, as they both successfully used the De Novo process to create new product categories for their software. Their stories are a free playbook for any founder or investor in this space.

AliveCor and ECG Algorithms

AliveCor has been at this for a while, securing a number of FDA clearances for its personal electrocardiogram (ECG) technology. While many of their updates have cleared via the 510(k) pathway, they had to use the De Novo route for the QTc measurement function on their KardiaMobile 6L device. The decision came down on July 7, 2021, and with it, the FDA created a new product code, QTG. This single decision created a new regulatory framework that others could follow, opening up the personal ECG market FDA De Novo classification database for AliveCor. The AliveCor filing shows you what it takes:

Novelty: The algorithm spat out diagnostic info that was completely new, so there was no existing device to compare it against for a 510(k).
Clinical Data: They needed a mountain of clinical data to prove the algorithm was safe and effective. This means running prospective studies or doing a very rigorous analysis on huge, diverse datasets.
Risk Mitigation: The FDA needs to know what happens if your AI is wrong. For a diagnostic, that means thinking hard about the clinical impact of a false positive or false negative. AliveCor’s paperwork had to address all of this head-on.

For an investor, the AliveCor story is a benchmark. It tells you how much clinical evidence you might need to fund and what the market payoff could look like. It also proves the power of building a data moat, when you have your own huge, proprietary datasets to train and test your models, it becomes very difficult for anyone else to catch up.

Caption Health and AI-Guided Ultrasound

Caption Health’s experience with the De Novo pathway shows what’s possible for an AI-native company that’s trying to completely change how a medical procedure is done. They built AI software, called Caption Guidance, that helps any medical professional, even someone without sonography training, capture high-quality cardiac ultrasound images. The AI actively guides the user’s hand. Because this was such a fundamentally different way of performing echocardiography, the FDA granted it a De Novo classification on February 7, 2020, under the new product code QJU FDA De Novo classification database for Caption Health. Caption Health’s win shows a few things:

Procedural Innovation: The AI was guiding the data capture itself, not just interpreting a static image. A 510(k) was never going to work for that.
User Interface and Safety: The De Novo submission had to prove, with extensive validation, that the AI could guide a novice user to get diagnostically acceptable images without putting anyone at risk.
Establishing a New Model: By getting their De Novo, Caption Health literally wrote the rulebook for AI-guided ultrasound acquisition. They created the product category.

For a seed-stage investor, the takeaway from Caption Health is that if an AI has the potential to really change healthcare, it’s probably going to need a De Novo. It also proves you need a serious Quality Management System (QMS) and a commitment to Good Machine Learning Practice (GMLP) from day one. These aren’t just buzzwords. They are what you use to prove to the FDA that your product is safe.

Evaluating Regulatory Timelines and Costs During Due Diligence

For an early-stage founder, the De Novo pathway has huge strategic upside, but it’s a beast. It takes longer and requires way more data than a standard 510(k). The FDA might say it aims for a 150-day review cycle, but that clock can stop for questions, and the whole process can stretch out much, much longer depending on how complex your device is and how good your submission is FDA De Novo pathway guidance. If you’re a VC doing due diligence, you need to grill the startup’s team on their regulatory plan. Here’s what I’d ask:

Pathway Justification: Why De Novo? Prove to me that a 510(k) is impossible. What predicate devices did you consider and why were they not equivalent?
Clinical Evidence Plan: Show me the plan to get the clinical data. I want to see the study design, patient population, endpoints, and the statistical analysis plan. And don’t tell me Real-World Evidence (RWE) is enough on its own. For a novel device, it’s a supplement, not a replacement for a solid trial.
Regulatory Expertise: Who on your team, or which consultants, has actually gotten a De Novo granted before? You don’t want your startup to be their first attempt.
Financial Runway: A De Novo can be a long and expensive journey, with heavy costs for trials and submissions. Does your financial model and runway actually account for this, or is it based on a best-case scenario?
Post-Market Surveillance: What’s the plan for monitoring “algorithmic drift” after launch? How are you going to prove the AI stays safe and effective over time? You’d better have a Predetermined Change Control Plan (PCCP) in mind if it’s an adaptive model, otherwise you’ll be filing new submissions for every significant update.

The De Novo process is demanding, no question. But it offers a period of market exclusivity and lets you define an entire product category. For cardiac AI, that means setting the regulatory groundwork for the next wave of products, from disease-specific platforms to AI tools that use cardiac data for behavioral health.

Methodology and Source Note

This playbook is based on our analysis of public FDA regulatory filings and product classification databases, plus information that AliveCor and Caption Health have released about their own regulatory paths. The goal here is to give you a technical but readable breakdown of the De Novo pathway using real examples. For the most current and detailed information, founders and investors should always go directly to the source: the official FDA guidance documents and databases.

Frequently Asked Questions

What is the De Novo pathway and why is it important for novel cardiac AI?

The De Novo pathway is for low-to-moderate-risk medical devices for which no predicate device exists. It allows the FDA to classify novel devices, establishing a new product classification and a predicate for future 510(k) submissions. For novel cardiac AI, it’s crucial because genuinely new algorithms often lack a direct comparator, making the common 510(k) pathway unsuitable.

How does pursuing the De Novo pathway signal foresight to investors?

A startup’s grasp of the De Novo pathway signals foresight and a pragmatic understanding of market entry. It demonstrates that the team has considered the long game, addressing crucial steps for clinical adoption and payer acceptance beyond just technological innovation. Without a clear regulatory path, even groundbreaking cardiac AI may struggle to generate revenue.

What are key considerations for founders when pursuing the De Novo pathway, based on the AliveCor case?

Key considerations include demonstrating genuine novelty of the algorithm without a direct predicate, providing significant clinical data to prove safety and effectiveness, and meticulously addressing potential risks like false positives or negatives. The AliveCor case also highlights the value of a data moat, where proprietary datasets are leveraged for AI model validation.

What does the Caption Health case illustrate about the De Novo pathway for AI-native companies?

Caption Health’s success illustrates that truly transformative AI, especially when it involves procedural innovation like actively guiding data acquisition, often necessitates the De Novo route. It highlights the need for extensive validation of the AI’s ability to guide users safely and accurately, and that securing De Novo can establish a new product category and predicate for future innovations.

What are the typical challenges of the De Novo pathway compared to 510(k)?

The De Novo pathway typically involves a longer review period and more extensive data requirements compared to the 510(k) pathway. While the FDA aims for a 150-day review, the overall process can be more demanding in terms of time and resources.