How Does a Medical Device Gain FDA Clearance?

cyber virtual fda

Even before most people were educated about what Covid-19 was, scientists began the important work of finding a cure. With unprecedented focus and resources to back these efforts, several effective vaccines were produced in less than a year. Anyone with experience in research and development of medical products understands the Herculean effort it took to accomplish this. A typical pathway to FDA approval of vaccines takes years, not months. The process of getting a medical device out on the market is also rigorous, but fortunately, the FDA provides helpful guidance to assist in the process.

The regulatory pathway

In theory, the pathway to government-sanctioned clearance looks like this:

Graphic credit: Medidee.com

In reality, for the company trying to gain approval for a product or solution, it can feel a bit more like this:

Any company seeking FDA approval has to be prepared for multiple setbacks and delays during the approval process.

So you’ve invented a cutting-edge new medical device! Good job! (Now what?)

Welcome to the pre-market phase of FDA clearance (or approval). First, you need to create a technical definition of your invention. What exactly are its intended purposes, the indications for use and intended user, and how does it work? Then you need to classify your device based on already existing classification examples organized into specialty “panels” and the identified risks. Your product will fall into one of three classifications: Class I, Class II, or Class III. Class I products pose a minimal potential for harm to the patient or user, whereas Class III devices pose the highest risks, presenting a high risk of illness or injury to the patient or end-user. Often, Class III devices can require implantation or are used to sustain life. Most medical devices are considered Class II, meaning they pose a moderate risk to patients and users.

Siim, Sahiner et al.
https://cdn.ymaws.com/siim.org/resource/resmgr/mimi18/presentations/18cmimi-sahiner.pdf

But what if your device really is cutting-edge, and there are no classification precedents (i.e. predicate devices) for it? You can submit a 513(g) Request for Information to the FDA for device determination—for a fee. If your device truly is cutting-edge and has no “substantially equivalent” predicated counterpart devices with a similar intended use and/or technical characteristics, the FDA may automatically categorize it as a Class III. If you believe this classification is inaccurate, then you have to initiate the DeNovo classification pathway process in order to apply for a downgrade in classification.

Medical device classification is determined by a wide range of different descriptors. For what field of medicine is the device intended? What is the scope of usage? (Is the device designed for adults? For children? For only one particular body part?) Is it hardware or software? If software, is it designed to detect or diagnose a pathology? Being concise and specific in the device’s intended use and indications for use descriptions, as well as its limitations for usage, can help in proper FDA classification of a medical device.

Pre-submission and the uncertainty of AI devices

Fortunately, the FDA provides a valuable opportunity to ask concrete questions and get helpful feedback for medical devices that may not have a clear substantially equivalent predicate device, the pre-submission (aka Q-submission). Medical software solutions, especially those incorporating AI technology, can be subject to a considerable amount of scrutiny because there are still many open questions by the FDA and industry as to how to treat AI medical devices.

Here is a chart with FDA codes for how computer-aided medical devices are categorized:

FDA, https://www.fda.gov/media/135712/download

Currently, AI-based medical devices cannot work autonomously; they can only be submitted as adjunct or supportive devices for the end user (such as a radiologist) and function to provide a second opinion. Currently, the FDA is working closely with manufacturers to develop best practices and standards for devices incorporating machine learning and artificial intelligence.

The graphic below shows which medical devices using AI have already been cleared by the FDA.

Benjamens et al. (2020)
https://www.nature.com/articles/s41746-020-00324-0

Proof of quality control in product development, validation, and risk analysis

The next steps in pre-submission are critically important, and it is crucial to get feedback from the FDA early on in order to stay on track:

  1. Product Development: Present a detailed description of product development process showing that standards were followed.
  2. Clinical Association: Conduct a systematic literature review and comparative analysis to prove that the device is state-of-the-art and relevant/useful to current medical practice.
  3. Analytical Validation: Using in-house data, demonstrate that the device performs correctly and reliably, and produces precise output data.
  4. Clinical Validation: In a clinical setting, conduct studies of the device to provide evidence that the device functions consistently as described and intended. 

An important thing to note about the entire development, verification and validation process is that everything must be traceable. Everything must be extensively documented, including precise metrics to evaluate the success of models and testing procedures and clear protocol to be followed for quality control. Through the entire clearance process, risk assessment and ethics considerations are important components.

One area of consideration in clinical validation studies is that ideally, the device should be tested on a wide range of patients that accurately represent the intended end-users. Providing the FDA with data showing that the device is safe and works as intended for people of varying ethnicity/race, sex, age, size, and weight is usually essential. It can be very time-consuming, expensive, and sometimes logistically challenging to engage in such widely diverse clinical studies. Additionally, data need to be interpreted by independent, unaffiliated evaluators to confirm that the device functions as well as or better than the current standard technology/methodology. Publication of peer-reviewed scientific papers is one of the final steps.

To sum up: The FDA regulatory pathway for medical device approval is a massive undertaking. Success requires careful planning from the beginning, with careful attention to the complexities of each step along the way. Fortunately, with the Q-submission, the FDA offers a program to receive early feedback to avoid being on the wrong track for too long.

Authors: Erika Greelish, André Meyer

Share on facebook
Share on twitter
Share on linkedin
Share on email

Related Posts

About Us
The ScanDiags software service, developed by Balzano Informatik AG, provides AI for augmented diagnosis from musculoskeletal MRI.

Let’s Socialize

Share on facebook
Share on twitter
Share on linkedin
Share on email
Popular Post