5 questions you need to answer to get your digital health or AI software product to US market quicker
Validate your digital health product from the word go
When it comes to taking health innovation to the US, digital health companies need to be armed with robust clinical evidence to achieve approval from the Food and Drug Administration (FDA). Don’t expect the FDA to approve your product merely based on your ambitious goals and lofty vision – evidence is key. Only products that have demonstrated impact – whether clinically superior or more cost-effective than existing solutions – should be entering the market. At Hardian, we have extensive experience in supporting companies to successfully navigate the US regulatory landscape. Over the years, we’ve noticed that startups often stumble over the same obstacles trying to get their product to the US market – particularly those developing AI or Software as a Medical Device (AIaMD/ SaMD) products. In this blog, we share 5 questions that you need to consider from the offset, that will maximise your chances of securing FDA regulatory approval and landing your product on US soil faster.
The adage “move fast and break things” may work for technology startups but rightfully has no place in healthcare. From the start, digital health companies should be thinking about evidence generation and clinical validation that matches their marketing claims. It is imperative to build a robust evidence base around the intended use of your digital health product before even considering entering the market.
Clinical research and validation sit at the intersection of product strategy and regulatory strategy. These 4 things are crucial, without having a clear clinical research and validation strategy all the other things won't work – they have to align together. Consider this from the start and trust us, you'll find that the FDA regulatory and go-to-market process isn’t as painful as a lot of startups find it.
Think of the FDA as your first customer
Considering the US market for your digital health technology? Start by thinking of the FDA as your first customer. Through the pre-submission process, the FDA helpfully provides digital health technology companies with formal feedback about their products prior to formal application submission. The pre-submission process offers startups an invaluable opportunity to directly seek advice from the FDA on particular areas of focus prior to full FDA submission. Even before building a product, startups should be considering how they will generate enough real-world data and evidence to demonstrate the value of their product and support regulatory approval. After all, you don't want to find yourself a couple of years down the line, having to re-design your product because you didn't consider how it would be regulated or the clinical validation required. We understand that building evidence that is robust and meaningful can be difficult, time-consuming, and expensive but with a bit of forward planning, startups can save a lot of time and energy.
5 questions for clinical validation you need to answer
1. What is the quickest, cheapest, and safest way to market?
Startups should be focusing on getting a safe, effective and cybersecure minimally viable product (MVP) to market quickly. Trying to achieve ambitious product goals from the outset can be a major stumbling block for startups, more complex products require more complex and expensive steps to achieve clinical validation. You can't claim your product can do A, B, C, D, E, and F, if you can only prove – with substantial evidence – that it can do A, B, and C. Start small and build from there.
2. What is the sample size needed to show robust validation?
How many people do you need (retrospectively or prospectively) to test your device or software on to fully validate it, and how will you justify the sample size? The FDA understands how difficult it is to get large sample sizes needed to develop AI-models. Still, it is important to ensure your sample size is generalisable across your target population – trialling your product on 5 people if it's going to be used on a broad range of ages just isn't going to cut it. Most importantly, you should be able to justify your sample size and demonstrate that the validation data is representative of the device’s intended use and adequate to draw meaningful conclusions.
3. What is your population dataset?
The FDA expects your validation to be representative of the US intended use population. What are the demographics of the sample? How old are they? Does the sample population have any comorbidities? Ensure these people are captured in the data set for your clinical investigation, as well as other factors to test adequate device performance, like different environments and device settings. Top tip: For FDA approval, at least 50% of your data set should be from the United States population.
Similarly, the FDA will want to see consideration paid to the user demographic, not just the patient demographic: education, training and experience of clinicians may not be the same in the SaMD manufacturer’s home market as it is in the USA; clinical workflows and clinical indicators may differ too, causing usability (human factors) problems when moving a device from one market to the other.
4. How have you set your acceptance criteria?
Acceptance criteria tells the regulators what you have set as a benchmark, and how you have set it - for example, the current standard of clinicians without being assisted by AI. This data can be sourced (accurately) from a systematic literature review which is essential to provide the pre-clinical and clinical data needed to support your application. Consider what the current best practice is for the field your product is intended for? How good is the current best practice, and how will you compare your product against this? There are three levels to consider of your SaMD: is it offering assistance or augmentation to the clinician, or is it autonomous i.e. taking the clinician out of the loop?
5. What are the equivalent devices and/or outputs?
If going for 510(k) clearance – approval of a device that is similar to one that has been previously approved by the FDA – you need to be able to demonstrate that your device or software is substantially equivalent to another device marketed legally in the US. In other words, the evidence you submit to the FDA doesn’t just show that the product is as effective and safe as a device that is already available, it also serves as additional evidence to support your case, for the use of your product on the target market (i.e. someone else has done this safely, so we can too). Ask yourself does your product perform the same function as the predicate device? Is your intended use the same? If so, then justify the similarities and show that the product is already on the market and functioning safely and effectively. Differences between the two devices are ok, as long the intended use is the same, and the nuanced differences do not raise new safety or efficacy concerns about your own product. However be cautious - you may be infringing on someone else’s intellectual property by telling the FDA you are equivalent to another patented device!
How can we support your clinical evidence generation?
With extensive experience across clinical research and evidence generation, we leverage our clinical and academic expertise to help you build a robust product with strong evidence. Regardless of the current stage, from hypothesis to implementation, we'll optimise clinical trial design to help you demonstrate the analytical and clinical validity of your digital health product.
Have you already received feedback from the FDA? By carefully and thoroughly analysing your product, its intended use, and your long-term goals, our experts will conduct a gap analysis – identifying the differences between the evidence you have now and what you need for FDA approval – to support your organisation to fill those gaps.
Hardian Health is a clinical digital consultancy focused on leveraging technology into healthcare markets through clinical strategy, scientific validation, regulation, health economics and intellectual property.
