As the biopharma space becomes increasingly competitive and the days of blockbuster drugs seem further behind, biopharma companies are investing in digital to find new and innovative ways to differentiate themselves. Digital leaders are turning to software as a medical device (SaMD) to help diagnose or treat disease.
These next-generation tools are creating more personalized experiences for patients, more information to guide clinical decision-making, and lots of data for biopharma companies to use to continuously improve therapies and justify reimbursement. Along with this growth have come several developments in the regulatory space that affect how SaMD products are classified and how they can be used.
According to the International Medical Device Regulators Forum (IMDRF), software as a medical device is software intended to be used for one or more medical purposes and is not part of a physical medical device. These purposes vary and may include assisting in the diagnosis or treatment of disease, assisting in symptom management, or as a diagnostic companion to a drug. There are many other applications as well.
SaMD software must be developed under a quality management system similar to hardware devices. There are standards and guidelines that apply specifically to SaMD. For example, IEC 62304 Software for medical devices — Software life cycle processes is an international standard that provides a framework for developing and maintaining SaMD and software within medical devices. There are several other standards and guidelines applicable to SaMD products.
Several trends surrounding the development of SaMD are impacting the way these products are classified and regulated, or pose challenges to these processes. Here are five of the top trends.
Trend 1: Widening of the divide between US and European regulations in the classification of a SaMD product
The rapid growth of the SaMD market is driving the expanded use of existing digital products in new ways that may change their classification as well as evolving classifications by regulators.
This makes it challenging for US biopharmaceutical companies to market a SaMD product if it does not fit easily into the rules-based classification system of the EU Medical Device Regulations or is not substantially equivalent to a US predicate device. These regulations are focused more on traditional medical devices, so software classifications may be more difficult to determine.
SaMD products may also contain software of unknown parentage or origin (SOUP). This is software that IEC 62304 defines as software that has not been developed to be incorporated into a medical device and/or where appropriate records of the development process are not available. Additional controls are therefore required to manage SOUP risk.
Trend 2: Certification of companies that develop SaMD, not every new SaMD product
Another important trend is how SaMD products are treated within the regulatory scheme. The FDA is working to pilot some programs that will help here. The idea is for the agency to provide a streamlined and efficient regulatory approach by pre-certifying the SaMD developer first, not primarily the product.
This would ease the ongoing monitoring burden on these companies and could reduce content submissions, speed up review, or both. The benefit comes from the fact that changes in software tend to happen much faster than in traditional hardware products.
Some of these products also use software-as-a-service models. In these cases, the software may be subject to continuous changes. The challenges are how do you manage these changes, keep the software compatible and ensure that the changes do not affect the functionality of the device?
Trend 3: AI and machine learning create a need for a framework that enables continuous change
The increased use of artificial intelligence and machine learning technologies has a significant impact on SaMD products. These advanced technologies allow the software to continuously “learn”. This is great for patient care, especially in cancer diagnosis and treatment. AI and ML are becoming an even bigger part of SaMD’s offerings. But this also means that these products are constantly changing.
The rules will need to keep up with these learning algorithms. It is important that algorithms continue to learn and perform advanced data processing to gain their full benefit. The traditional paradigm requires the learning algorithm to be “locked” and would likely require the FDA to conduct a premarket review for any changes.
A new regulatory approach could allow for a continuous learning environment where the algorithm can continue to learn and change, but also have adequate regulatory oversight to ensure that safety and efficacy are maintained. While under discussion by the FDA, this model does not yet exist.
Trend 4: Launching devices that have no predicate examples
The growing experience for many organizations is that a new SaMD product they want to bring to market is completely new or unique. If a pharmaceutical company plans to use the traditional FDA 510(k) route to market, it usually requires the company to have a predicate device or substantial equivalent.
In many cases, there is no predicate device available, so the FDA 510(k) pathway to market clearance cannot be used. The organization must classify the device through a De Novo classification request or go through FDA preapproval, which is more rigorous, time-consuming, and expensive.
Trend 5: Urgent need for successful partnerships with third parties
Another growing trend is the need for pharmaceutical companies to have trusted third-party partners who can help build and then bring SaMD products to market, as well as help navigate the complex regulatory path along the way.
The ideal partner is one that knows the current regulatory climate, but also knows what is being worked on that will impact future product development. Such a partner can build the products within a medical device quality management system, bring the product to market, and handle regulatory hurdles on behalf of the customer.
Some third-party vendors may be able to help with software development, but a truly valuable partner is one who can help manage risk during the development cycle and address any other regulatory issues or concerns, with that a SaMD product may encounter on its way from drawing board to market.
Photo: Elena Lukyanova, Getty Images