Software as a Medical Device
Second Edition
1 Regulatory Affairs Professionals Society
The current world, where almost all medical devices and
in vitro diagnostic (IVD) devices are connected, is chang-
ing more rapidly than ever before, with the rapid onset
of artificial intelligence (AI) in virtually all aspects of our
modern-day society. The new horizon features AI-enhanced
products, AI-supported assessment and decision tools, and
various AI products that, in one way or another, support the
healthcare system and providers, allowing, for example, longer
periods of independent and assisted living. Soon after its
global public launch, the fast-developing technology cannot
be disconnected from our healthcare world as we enter the
second quarter of the 21st century.
Today’s connectivity of medical devices enables manufac-
turers to migrate essential and supporting functionality from
the device’s computing resources to personal computers, hand-
held devices, cloud servers, wearables, and other products in the
Internet of Things. Those external computing platforms often
offer more powerful capabilities and are easier to use, maintain,
scale, and upgrade. Such medical device software enhances
the original medical device’s capabilities and performance. The
onset of AI in computing and data analysis brings opportu-
nities but also initially poses significant regulatory challenges,
which the regulatory world is working hard to overcome.
As manufacturers have centralized almost everything
that can be centralized, we are starting to see the opposite
trend. As software gets more powerful and data-hungry, it
clogs the data pipelines. At the same time, its connectiv-
ity may pose privacy, general, and cybersecurity challenges.
Software manufacturers are now increasingly offloading
functionality, computation, and data storage onto the spare
computing resources of medical devices and wearables, close
to where the raw data is generated. Such distributed com-
puting is known as edge computing. Decentralized machine
learning will further amplify this trend. The wider use of AI,
beyond its subset of machine learning, is further accelerating
the speed of change and innovation.
Manufacturers’ flexibility in choosing where computing
occurs and in how they bundle and sell software functionality
or assign medical claims creates a certain degree of device flu-
idity. This trend will further intensify as continuous machine
learning and broader AI support enhance the capabilities
of medical devices. Alignment and further interpretative
guidance are needed to help us all continue on our way to
ensure such devices are safe and perform as intended. As data
is getting central in the regulatory debate, AI-enabled devices
ride on the crest of the wave of finding real world evidence,
typically more pronounced after market launch. Postmarket
is the new truth serum – where lifecycle management turns
real-world data into a continuous evidence engine.
The International Medical Device Regulators Forum1
created the term Software as a Medical Device (SaMD) for
regulatory purposes. The IMDRF work intended to enable
a uniform language, common definitions, and regulatory
concepts. It has partially achieved those goals, prompting
jurisdictions to adjust general concepts to SaMD’s specific
needs. Nevertheless, definitions vary somewhat, and some
jurisdictions have added specific terminology. Also, of course,
nowadays wearables take flight. In Chapter 2, Pat Baird, Koen
Cobbaert, and Zhuo Li explain the foundations of SaMD,
share their views on the modularization of medical device
functionality, and discuss the regulatory status of wearables.
As software and, in particular, AI is eating the medical
device world, the in-vitro diagnostics (IVD) arena follows
1
Introduction
Koen Cobbaert, MSc, and Gert Bos, PhD, MSc, FRAPS
Postmarket is the new truth serum
– where lifecycle management
turns real-world data into a
continuous evidence engine.
Second Edition
1 Regulatory Affairs Professionals Society
The current world, where almost all medical devices and
in vitro diagnostic (IVD) devices are connected, is chang-
ing more rapidly than ever before, with the rapid onset
of artificial intelligence (AI) in virtually all aspects of our
modern-day society. The new horizon features AI-enhanced
products, AI-supported assessment and decision tools, and
various AI products that, in one way or another, support the
healthcare system and providers, allowing, for example, longer
periods of independent and assisted living. Soon after its
global public launch, the fast-developing technology cannot
be disconnected from our healthcare world as we enter the
second quarter of the 21st century.
Today’s connectivity of medical devices enables manufac-
turers to migrate essential and supporting functionality from
the device’s computing resources to personal computers, hand-
held devices, cloud servers, wearables, and other products in the
Internet of Things. Those external computing platforms often
offer more powerful capabilities and are easier to use, maintain,
scale, and upgrade. Such medical device software enhances
the original medical device’s capabilities and performance. The
onset of AI in computing and data analysis brings opportu-
nities but also initially poses significant regulatory challenges,
which the regulatory world is working hard to overcome.
As manufacturers have centralized almost everything
that can be centralized, we are starting to see the opposite
trend. As software gets more powerful and data-hungry, it
clogs the data pipelines. At the same time, its connectiv-
ity may pose privacy, general, and cybersecurity challenges.
Software manufacturers are now increasingly offloading
functionality, computation, and data storage onto the spare
computing resources of medical devices and wearables, close
to where the raw data is generated. Such distributed com-
puting is known as edge computing. Decentralized machine
learning will further amplify this trend. The wider use of AI,
beyond its subset of machine learning, is further accelerating
the speed of change and innovation.
Manufacturers’ flexibility in choosing where computing
occurs and in how they bundle and sell software functionality
or assign medical claims creates a certain degree of device flu-
idity. This trend will further intensify as continuous machine
learning and broader AI support enhance the capabilities
of medical devices. Alignment and further interpretative
guidance are needed to help us all continue on our way to
ensure such devices are safe and perform as intended. As data
is getting central in the regulatory debate, AI-enabled devices
ride on the crest of the wave of finding real world evidence,
typically more pronounced after market launch. Postmarket
is the new truth serum – where lifecycle management turns
real-world data into a continuous evidence engine.
The International Medical Device Regulators Forum1
created the term Software as a Medical Device (SaMD) for
regulatory purposes. The IMDRF work intended to enable
a uniform language, common definitions, and regulatory
concepts. It has partially achieved those goals, prompting
jurisdictions to adjust general concepts to SaMD’s specific
needs. Nevertheless, definitions vary somewhat, and some
jurisdictions have added specific terminology. Also, of course,
nowadays wearables take flight. In Chapter 2, Pat Baird, Koen
Cobbaert, and Zhuo Li explain the foundations of SaMD,
share their views on the modularization of medical device
functionality, and discuss the regulatory status of wearables.
As software and, in particular, AI is eating the medical
device world, the in-vitro diagnostics (IVD) arena follows
1
Introduction
Koen Cobbaert, MSc, and Gert Bos, PhD, MSc, FRAPS
Postmarket is the new truth serum
– where lifecycle management
turns real-world data into a
continuous evidence engine.