As each day passes, trends like artificial intelligence (AI), cloud computing, machine learning and the Internet of Things (IoT) creep a little closer into our daily lives. While the public at large may not be fully aware of the implications of these developments, industry experts realize the profound impact that such technologies will have on dozens of industries, such as medical care. In the following Q&A, Arvind Ananthan, Medical Devices Industry Manager, MathWorks, discusses how AI, IoT, and cloud computing are revolutionizing the way doctors interact with patients before, during, and after the point-of-care to render more effective treatment and foster informed doctor-patient relationships.
Embedded Computing Design: Emerging technology trends such as AI and IoT are driving a new era of healthcare. What technologies in particular are spurring innovation in the medical industry?
ANANTHAN: While the effects of technologies such as AI and IoT may seem to be more hype than substance, the changes emerging from neural networks, cloud computing, machine learning, deep learning, and wearables are quite real.
Just look at the enormous amount of data that currently available to the traditional healthcare provider. Patient records, medical imaging and video files, and vital sign monitors are just a few data sources with deep potential for improving patient care. For example, predictive analytics systems can collect and store medical data at the point-of-care, allowing engineers and scientists to find trends that improve hospital procedures and patient outcomes.
Embedded Computing Design: If the influx of medical data creates opportunities, does it also introduce new challenges?
ANANTHAN: As AI systems, wearables, and other new technologies move from concept to reality, a few challenges have remained consistent. One is the need to aggregate data that is scattered across multiple platforms, apply the latest analytics capabilities, and transform the data into something that’s actionable. Engineering software tools like MATLAB are helpful in that they allow medical device engineers and researchers to analyze large amounts of varied data types and then quickly begin prototyping and algorithm implementation. And they can develop and deploy new machine-learning models without coding them from scratch, which reduces time-to-market.
Other significant challenges on the regulatory front also remain for developers in this space, especially for the non-incumbents entering this domain anew. We are also seeing some proactive changes from the FDA (Software Pre-certification program) in the area of regulating digital health apps and products in a more transparent way that reduces the burden on innovators.
Embedded Computing Design: What other considerations should designers be mindful of when applying software tools to their projects?
ANANTHAN: Deducing trends and uncovering insights can be extremely difficult when faced with an overwhelming amount of scattered, seemingly random data points spread across various datatype domains (signals, images, video, tables). To ensure we are applying the correct predictive analytics system, we must first start by breaking down the system framework behind the specific technology we’re pulling information from.
There are two central perspectives that simplify this emerging landscape: the first is the IoT system framework and its underlying infrastructure; and the second is the data analytics (machine learning) framework, which focuses on the smart algorithms that help physicians and patients make more informed, data-driven decisions, for example.
Embedded Computing Design: Can you elaborate on the IoT system framework? What are the central elements?
ANANTHAN: The IoT system framework is one of two industry views that can help us to characterize digital health. It is comprised of three main elements: edge nodes, gateway, or cloud aggregators, and a back-end data analytics engine that operates on aggregated data for trend analysis, anomaly detection, etc.
Embedded Computing Design: What is the importance of an edge node? How does it work in conjunction with the full IoT system framework?
ANANTHAN: Edge nodes are the first step in the IoT system framework, and they are responsible for collecting raw physiological health data through various sensors. In an IoT system, edge nodes can be anything from wearable fitness devices to blood glucose sensors, ECG monitors, or even a smart pill that transmits data to a receiver worn by patients. After the data has been collected from the edge node, it is processed through a specific signal or algorithms that extracts and transmits signal features to a cloud aggregator.
Embedded Computing Design: What are the benefits of pairing the signal or processing algorithm with sensor data streaming from an edge node?
ANANTHAN: Pairing the correct signal or processing algorithm with specific edge node data leads to direct benefits for cloud aggregators, including reduced bandwidth requirements and improved power efficiency for a specific wearable by intelligently pre-processing the data and limiting the data transferred back to the gateway and cloud. Furthermore, improved wearable efficiency leads to reduced device size and longer battery life.
Processing algorithms, specifically feature extraction algorithms, also enable compressed information to be directly funneled to the cloud and aggregated for a single patient or across a broader patient population instead of piping the entire raw data stream. Ultimately, predictive analytics can be applied to this larger data set to provide patients and physicians with real-time reports.
However, engineers and scientists must realize that the right combination of algorithms for preprocessing and feature extraction is a critical, but difficult, step in this workflow – critical because the combination will determine the effectiveness of the final predictive analytics solution, and difficult because it can be time-consuming to figure out the appropriate partitioning of algorithms without proper prototyping tools. Another important consideration is the field configurability of the algorithms – cloud analytics can be updated as needed at any time, while the algorithms implemented on the wearables may be less modifiable depending on the architecture or hardware chosen.
Embedded Computing Design: Can you elaborate on the data analytics system framework? What are the central elements?
ANANTHAN: The data analytics framework, where machine learning algorithms play a central role, adds intelligence to the digital health system and enables the extraction of meaningful information from large amounts of text data, signals, images, and videos to automate and accelerate diagnostic capabilities. Like the IoT system framework, the data analytics framework is comprised of three central components: pre-processing of data, feature extraction, and development of predictive models.
Embedded Computing Design: Do the IoT and data analytics system frameworks operate in separate silos or is there a level of integration and/or overlap?
ANANTHAN: In order for digital health systems to operate efficiently and offer the biggest benefit to patients and physicians, there must be a certain level of integration between IoT and data analytics systems. Typically, an IoT system will enable the edge node to collect signals and data coming from the device to begin prototyping signal processing algorithms, while the data analytics system explores various machine-learning models and rapidly implements them on a target platform through C-code generation technologies. Although this is a complex relationship, it allows companies to push the boundaries of what’s possible in digital health.
Embedded Computing Design: While emerging digital technologies will undoubtedly continue to play a pivotal role in the world of medicine, what do you consider to be the top benefits from these new trends, specifically predictive analytics?
ANANTHAN: At a broad level, these digital technologies will help move healthcare regimes toward data-driven personalized medicine. In fact, it’s quite possible that predictive analytics will drive the progress of both preventative and therapeutic care, with data collected from wearables and shared on personal devices that has the potential to enable more innovative therapies (such as electroceuticals) in addition to the pharmaceutical compound-based therapies we have today. Predictive analytics will also allow for a more informed and personal relationship between patients and physicians, leading to more effective diagnoses and early treatment of a wider range of physiological conditions.
The MathWorks, Inc.
Google+: www.facebook.com/MATLABeletter-03-28-2018 eletter-03-29-2018
About the Author
Richard Nass is the Executive Vice-President of OpenSystems Media. His key responsibilities include setting the direction for all aspects of OpenSystems Media’s Embedded and IoT product portfolios, including web sites, e-newsletters, print and digital magazines, and various other digital and print activities. He was instrumental in developing the company's on-line educational portal, Embedded University. Previously, Nass was the Brand Director for UBM’s award-winning Design News property. Prior to that, he led the content team for UBM Canon’s Medical Devices Group, as well all custom properties and events in the U.S., Europe, and Asia. Nass has been in the engineering OEM industry for more than 25 years. In prior stints, he led the Content Team at EE Times, handling the Embedded and Custom groups and the TechOnline DesignLine network of design engineering web sites. Nass holds a BSEE degree from the New Jersey Institute of Technology.Follow on Twitter More Content by Rich Nass