Software as a Medical Device: Key Concepts and Why It Matters in Healthcare

Overview

New advancements in connectivity, artificial intelligence, and image storage and processing have led to the development of new tools for healthcare providers, including software as a medical device (SaMD). Though the promise of this technology is vast, with seemingly unlimited potential to detect and combat disease, as with most technologies, compliance with regulatory bodies is paramount to ensure patient safety and protect sensitive data.

What is SaMD?

The technological advances of the past decade have launched the world into a “Fourth Industrial Revolution,” epitomized by the proliferation of interconnected devices that can communicate with one another across the globe and interface with humans in new and exciting ways [1]. Among these advancements is the rise of specialized healthcare software that may involve the “diagnosis, prevention, monitoring, treatment, or alleviation of disease” [2]. When the software is “intended to be used for one or more medical purposes that perform these purposes without being part of a hardware medical device,” it is classified as SaMD [3]. Although a medical device is technically any item that impacts healthcare, from a tongue depressor to an MRI, there is a clear delineation between what is and is not SaMD. For example, software that operates a pacemaker or controls an infusion pump would not be considered SaMD, nor would systems that provide the foundation for maintenance, storage, updating, and accessing electronic medical records [4].

Why is SaMD important in healthcare?

The goal of any healthcare technology should be to create positive outcomes for patients and advance public health initiatives. To this end, SaMD can effect change in numerous ways, such as identifying and flagging trends across vast swaths of electronic medical records to identify potential health risks [5]. It can also take advantage of the ease with which digital systems can synthesize massive troves of data to uncover patterns and make predictions. For example, the confluence of wearable medical devices, the internet (for connectivity and secure cloud storage), and artificial intelligence provides an opportunity for SaMD to analyze, interpret, and learn from a wealth of real-time data in ways that would be impossible for a single human physician. The result of these advancements is the creation of SaMD that can collect and store information from a host of factors related to disease onset, progression, and treatment, including [6]:

• Genetic and epigenetic factors
• Demographic variables
• Environmental and contextual factors
• Lifestyle patterns

SaMD that utilizes artificial intelligence and/or machine learning algorithms can be used to enhance diagnostic accuracy and provide earlier disease detection than is currently possible. For example, SaMD is currently being evaluated for efficacy in detecting diabetic retinopathy (DR) [7]. It has also been used to identify various skin cancers, with studies showing that some SaMD performs comparable to, if not better, than trained clinicians and dermatologists [8].

The Regulatory Context for SaMD

Though SaMD holds great promise, it is incumbent upon government regulators to establish the necessary policies and procedures to balance innovation with patient safety. For example, in the United States, the Food and Drug Administration (FDA) evaluates risks and outcomes before granting approval for a device. Before SaMD may be used within the European Union (EU), it must receive a CE (“conformité européenne”) conformance mark. Within the People’s Republic of China, SaMD must first undergo technical review by the Chinese Center for the Evaluation of Medical Devices (CMDE), with final approval granted by the National Medical Products Administration (NMPA) [9]. In order to “harmonize the regulatory requirements for medical products that vary from country to country” [10], the International Medical Device Regulators Forum (IMDRF), which counts the U.S., EU, and China among its members, has produced a number of key documents, for topics ranging from clinical evaluation and risk categorization to how to apply a quality management system [11]. Thus, though approval must be obtained within specific countries, the IMRDF ensures that the regulatory guidelines that SaMD developers must meet have some degree of uniformity.

Challenges and Opportunities for SaMD Deployment

In addition to navigating legal channels and complying with the regulations, there are other challenges associated with deploying SaMD. Because most SaMD harnesses the power of connectivity (in other words, via the internet), there is a need for a standardized infrastructure and interoperability that allows information to flow freely between systems from different manufacturers, between medical disciplines, and across borders of nation-states [12]. However, such connectivity and storage of medical data carries inherent risk, which must be accounted for with enhanced security and encryption protocols, contingency plans for backup, and protocols for worst-case scenarios, such as denial of service attacks or cyberattacks [13].

Summary

The Fourth Industrial Revolution’s focus on developing cyber-physical systems has the potential to revolutionize healthcare. Better patient outcomes, achieved through medical personalization and diagnostic advances, could positively impact the lives and longevity of millions of people, provided the technologies are developed with a consideration of potential risks and comply with an aligned set of regulations being put into place across the globe.

View the article on our blog for a complete list of references:

• Essentials of Software as a Medical Device & Clinical Decision Support Systems
• Technology, Ethics, and Regulations
• EU AI Act
• Biotility: Regulatory Approval for Medical Devices (United States)
• Essentials of Responsible AI

This course was authored by:

• Informed Consent and Research with Wearable Tech
• Bring Your Own Device (BYOD) Studies
• The Playbook: Successfully Developing and Deploying Digital Clinical Measures

This webinar was presented by:

• On Tech Ethics Podcast – Synthetic Data in Research and Healthcare
• On Research Podcast – Artificial Intelligence in Research
• On Tech Ethics Podcast – Governance of Genetic Information

• Understand the Difference Between FDA-Cleared and FDA-Approved Medical Devices
• The Foundation for Good Machine Learning Practice (GMLP)
• Uses of AI in Drug Development
• Guidance on Managing Investigational Products in DCTs

• Assistant Director of the IACUC and IBC
• Research Finance Budget Analyst
• Intake Coordinator
• IRB Analyst
• Human Research Protection Program (HRPP) Analyst

Visit CITI Program’s Employment Center to view all open positions.