The impact of big data, artificial intelligence, and machine learning on the medical technology industry

The core business of medical technology companies (MedTechs) has been manufacturing and marketing physical devices Physical devices will continue to be a substantial part of their business, but on their own, are unlikely to deliver high growth rates, which are more likely to come from artificial intelligence (AI) data driven strategies that improve patient outcomes   The impact of big data, artificial intelligence, and machine learning on the medical technology industry   James Carville, an American strategist, who played a leading role in Bill Clinton winning the 1992 presidential race, insisted that the campaign focus on the economy and coined the phrase “It’s the economy, stupid”. If Carville was asked today for a winning long-term growth strategy for medical technology companies, might he say, “It’s big data, stupid”?   This Commentary suggests that while physical products have been the backbone of MedTech companies in the past, they are unlikely to contribute significantly to future growth rates, which are more likely to come from artificial intelligence (AI) driven big data innovations, which create new solutions that improve patient journeys and outcomes.   In this Commentary   This Commentary describes the meaning of ‘big data’ in a healthcare context, explains ‘the data universe’ and stresses not only its immense volume, but also its variety, and the phenomenal speed at which the data universe is growing. Today, most industries leverage big data and AI techniques to create innovative offerings that drive growth and enhance competitive advantage. However, with few exceptions, traditional MedTechs have been relatively slow to collect and analyse a wide range of health, medical and lifestyle data which have the potential to provide innovative software offerings that improve patients’ therapeutic journeys and complement physical products. This is partly because the industry must adhere to strict regulations and partly because many medical technology companies lack the necessary capabilities and mindsets to collect and leverage big data. Most have business models that tweak legacy physical products and accept growth rates of ~5% as the ‘new normal’. We provide a brief history of big data and AI business strategies mainly to underline that these are relatively new. It was only in the early 2000s that electronic health records (EHR) began to replace paper-based patient records, which were stored in numerous filing cabinets in healthcare silos. It was not until ~2015 that EHRs became standard practice and researchers started to apply algorithms to EHRs and other data to detect patterns and make predictions that could improve diagnoses and treatments, enhance patient outcomes, and reduce healthcare costs. The increased use of big data and AI techniques in healthcare raises important cybersecurity concerns and trust issues because health professionals and patients do not understand how algorithms arrive at their conclusions and actions. Cybersecurity concerns are addresses by a range of encryption techniques and security protocols. Trust in algorithms has been helped by the development of  ‘explainable AI’, which is software that describes the essence of algorithms in easily understood terms. However, more work is still needed in these two areas. We introduce cloud and cloud services together with an explanation why these have experienced such rapid growth across all industries in recent years. The cloud makes it easier to store and access big data via the internet from anywhere in the world. Cloud services provide security for big data as well as a range of management and analytical tools that help to transform data into revenue generating software offerings. For MedTech companies, the cloud and cloud services provide the basis for more efficacious R&D. The medical technology industry has become bifurcated between companies that leverage AI driven big data strategies to enhance growth rates and those that predominantly focus on legacy physical product offerings and settle for lower growth rates. Over the past decade the nature of the medical technology industry has changed; partly because of AI big data strategies supported by the cloud computing and a large and rapidly growing range of open-source, easy-to-use AI tools. This has given small companies a competitive advantage. The Commentary concludes by describing a few of these small MedTechs with disruptive digital products that target large, rapidly growing, underserved market segments.          Big data and healthcare Big data are comprised of a wide range of information collected from multiple sources that surpasses the traditionally used amount of storage, processing, and analytical power and is unmanageable using conventional software tools. In healthcare settings, big data include hospital records, medical records of patients, results of medical examinations, and data generated by traditional medical devices as well as various biomedical and healthcare tools such as genomics, wearable biometric sensors, and smartphone apps. Biomedical research also generates data relevant for the medical technology industry.   The data universe The massive amount of data, which is generated from the entirety of the internet is referred to as the ‘data universe’. It is not only its volume that makes this special, but it is also the variety of the data and the phenomenal speed at which the universe is growing. The International Data Corporation (IDC) estimated that the data universe grew from ~130 exabytes in 2005 to >40,000 exabytes in 2020.  To put this in perspective: 1 gigabyte of data is 1bn bytes (18 zeros after the 1 or 230 bytes), and 1 exabyte is equal to 1bn gigabytes.

Data generated healthcare innovations In the past, collecting and interpreting vast quantities of data was not feasible, partly because computer systems were relatively small and did not generate much data, and partly because technologies to manage big data were underdeveloped. Fast forward to the present, and businesses across most industries now generate enormous amounts of data. Organizations apply AI and machine learning (ML) techniques to these data to create innovative product offerings to access new revenue streams with significant growth potential. Such technologies, combined with health-related big data, can positively impact the medical technology industry by generating novel diagnostics and treatments for patients, streamlining the process of medical record keeping and developing more personalized and responsive care plans that improve patient journeys and outcomes. You might also like:    Is the digital transformation of MedTech companies a choice or a necessity?   Leaning-in on digital and AI

The new rapidly evolving AI data driven healthcare ecosystem Despite the potential commercial advantages of AI data driven diagnostic and therapeutic solutions, many traditional MedTechs have been slow to collect health and lifestyle data from multiple sources to develop software offerings, which complement their legacy physical products. One notable exception is Philips Healthcare. In the early 2000s, the company was challenged by new entrants to the market who were successfully leveraging information from health wearables and other sources to create and market AI data driven offerings. At the 2016 annual conference of the American Healthcare Information and Management Systems Society (HIMSS) in Chicago, Jeroen Tas, a Philips executive, said, “We are in the midst of one of the most challenging times in healthcare history, facing growing and aging populations, the rise of chronic diseases, global resource constraints, and the transition to value-based care. These challenges demand connected health IT solutions that integrate, collect, combine, and deliver quality data for actionable insights to help improve patient outcomes, reduce costs, and improve access to quality care”.   Philips had the mindset and resources to respond positively to this rapidly changing ecosystem. In 2017 the company appointed Tas as its Chief Innovation & Strategy Officer, tasked with launching a suite of big data AI driven solutions, the IntelliVue® patient monitors, which support the growing demands of health professionals to provide quality care and improved outcomes for an expanding population of older, sicker patients with fewer resources. These monitoring solutions seamlessly connect big data, AI technology and patients to support health professionals to manage patients as they transition through their care journeys. In 2016, Philips and Masimo, a medical technology company specializing in non-invasive AI data driven patient monitoring devices, entered a multi-year business partnership involving both companies’ innovations in patient monitoring. Philips agreed to integrate Masimo's measurement technologies into its IntelliVue® monitors, to help clinicians assess patients’ cerebral oximetry and ventilation status. The outcome of the collaboration was the launch of a new suite of patient solutions, called Connected Care, which give healthcare providers the ability to monitor patients more effectively and reduce costs.   The bifurcation of the MedTech market In addition to large MedTechs such as Philips and Masimo, there are hundreds of small companies developing AI driven big data offerings aimed at improving patient outcomes. The reasons for many traditional companies’ slowness to fully leverage big data and AI applications are partly because medical devices are required to comply with stringent regulatory guidelines and partly because of the lack of capabilities. The different responses have bifurcated the industry. On the one hand there are traditional MedTechs, which predominantly focus on existing customers and market legacy physical offerings in slow growing markets. On the other hand, there are many small companies and a few very large medical technology corporations, which have embraced AI driven big data patient-centric solutions.   A brief history Big data has its genesis in the 1950s and 1960s when scientists and mathematicians began exploring the possibility of using computers to process large amounts of data to make intelligent decisions. This led to the development of technologies such as the first neural networks, which laid the foundation for modern Deep Learning. In the 1980s, researchers at IBM popularized the concept of big data to describe the process of collecting and analyzing large amounts of data, which empowered organizations to gain insights from information that previously was too complex to process. The 1990s saw the development of AI and ML, which enabled computers to learn from data and make decisions without the need for explicit programming. By the early 2000s, AI-based algorithms empowered machines to learn from data and make predictions. Many organizations, across a range of industries, saw the commercial opportunities of this and acquired capabilities to collect, store and analyse large amounts of information to identify patterns and trends that were previously impossible to detect.  Without large amounts of data, AI and ML techniques are less effective, which is significant for healthcare and the medical technology industry.   Big data in healthcare AI driven big data strategies are becoming increasingly important in healthcare. This is because AI techniques applied to masses of health-related information can improve patient care, enable more effective decision-making, reduce costs, identify new treatments, explore new markets, and create more efficient healthcare systems. Further, such applications can provide more accurate and timely diagnoses, as well as insights into how various treatments affect different people. As increasing amounts of health information become available, and data handling techniques improve, so traditional MedTech companies will have opportunities to boost their growth by complementing their physical devices and volume-based care with digital assets and personalised care.   Paper-based mindset Until recently health professionals were responsible for most of the different types of data associated with a patient’s treatment journey, which included medical histories, known allergies, medical and clinical narratives, images, laboratory examinations, and other private and personal information. Until the early 2000s these data were recorded on paper and stored in filing cabinets across numerous healthcare departments. It was not until 2003 that the US Institute of Medicine used the term ‘electronic health records’ (EHR). By 2008, only ~10% of US hospitals were using EHRs, which increased to ~80% by 2015. As EHRs became standard practice across multiple providers and data interoperability issues were resolved, the provision of healthcare improved, and medical errors and healthcare costs were reduced. Currently, the American National Institutes of Health (NIH) is inviting 1m people from diverse backgrounds across the US to help build a comprehensive big data set, which can be used to learn more about how biology, environment and lifestyles affect health in the expectation of discovering new ways to treat and prevent disease.   Trust and medical algorithms   As AI driven big data applications have increased, so trust in algorithms has been raised as an issue. This has been a major concern in healthcare. To address this challenge, explainable AI, has been developed. This is an AI technology that explains decisions and actions made by algorithms in a way that is easily understood by health professionals and patients. Explainable AI has helped to create trust in algorithms by providing a level of transparency, understanding and accountability. Further, incorporating feedback from medical professionals, patients, and other stakeholders into the development of medical algorithms has also helped to build trust. However, this entails collecting a wider variety of data than many healthcare companies are used to.

You might also like: Can elephants be taught to dance? Have diversified medical technology companies blown their competitive advantage? Big data healthcare strategies and security   With the increasing number of big data and AI healthcare solutions, cybersecurity has become a concern. Reducing this involves using technologies such as data encryption, secure cloud computing (see below), and authorization protocols to protect data stored in large databases. Additionally, organizations may use AI-driven applications to monitor their systems to find anomalies, detect malicious activity and unauthorized access to sensitive, personal information. To ensure the security of healthcare data, organizations also employ measures such as risk assessments, incident response plans, and regular security training of their staff.

Cloud storage and services Since the early 1990s, big data have benefitted from cloud storage, which makes it easier to store and access data over the internet and helps businesses to become more efficient and productive. It also offers organizations scalability, more control over their data and reduced costs. Organizations can: (i) easily increase their storage capacity as their data needs grow, (ii) access their data from anywhere in the world, and (iii) stop investing in expensive local storage devices. Further, cloud storage is becoming more secure, with encryption and other security measures making it safer to store data.   Companies moving their data from local storage devices to the cloud is more than just a simple transfer process and can be a complex, multi-year journey. Any organization that has accumulated several legacy databases and infrastructures will have to develop and manage a hybrid architecture to transfer the data. However, once in place and shared among stakeholders, cloud-based platforms can assist in unlocking clinical and operational insights at scale while speeding up innovation cycles for continuous value delivery. In combination with a secure and interoperable network of connections to hospital systems, cloud-based solutions represent an opportunity for healthcare leaders to unlock the value of data generated along the entire patient journey, from the hospital to the home. By turning data into insights at scale, it is possible to empower healthcare professionals by helping them to deliver personalized care, improved patient outcomes and lower costs.   The cloud also offers an increasing number of computing services. These are provided by companies such as Amazon Web Services, Google Cloud Platform, Microsoft Azure, IBM Cloud, Oracle Cloud, and Rackspace Cloud. The services include: (i) Infrastructure-as-a-Service (IaaS), which provides users with access to networks, storage, and computing resources, (ii) Platform-as-a-Service (PaaS) helps users to develop, run, and control applications without the need to manage infrastructure, (iii) Software-as-a-Service (SaaS), provides access to a variety of applications, (iv) Data-as-a-Service (DBaaS), gives users access to several types of databases, and (v) Serverless Computing enables users to run code without needing to provision or manage servers. Such services are expected to continue growing and help to transform healthcare. The provision of cloud computing services in healthcare makes medical record-sharing easier and safer, automates backend operations and facilitates the creation and maintenance of telehealth apps. The increasing use of data and cloud services by MedTech companies helps to break down data silos and develop evidence-based personalized solutions for a connected patient journey. In 2020, the healthcare cloud computing market was valued at ~US$24bn, and it is expected to reach ~US$52bn by 2026, registering a CAGR of >14% during the forecast period. Major drivers of cloud services include the increasing significance of AI driven big data applications.   Changes the nature of R&D Further, the cloud can change and speed up R&D. The starting point for MedTech R&D should be evolving patient needs and affordability. Healthcare-compliant cloud platforms offer a flexible foundation for the rapid development and testing of AI driven big data solutions created by cross functional teams working across an entire life cycle of an application: from development and testing to deployment. This changes medical technology companies’ traditional approach to R&D by transforming it into short cycles undertaken by multiple stakeholders. This modus operandi is replacing traditional lengthy and expensive R&D often carried out in an organisational silo and constrained by annual budgeting cycles. This often means that a significant length of time passes before an innovation gets into the hands of health professionals and patients for testing. Digital health solutions, on the other hand, can be tested by physicians and patients early in their development and improved features quickly added.      Free and easy to use AI and ML software libraries In the early 2000s, when AI and ML were in their infancy, companies needed data engineers with advanced mathematical capabilities to build complex AI systems. Today, this is unnecessary because of the development of simplified AI and ML libraries such as PyTorch and Tensorflow. These are free, easy to use, open-source, scalable AI, and ML packages, which reduce the need for data engineers to have advanced mathematical skills to build effective software health solutions. PyTorch, released in 2016,  was developed by Facebook and then Meta AI, and is now part of the Linux Foundation. The technology is known for its ease of use and flexibility, making it favoured by developers who want to rapidly prototype and experiment with new ideas. Its tools support graphics processing, which is popular with deep learning medical imaging strategies that involve training large, complex models on big data. TensorFlow was developed by the Google Brain team and originally released in 2015 for internal use.  It is a highly scalable library for numerical computations and allows its users to build, train and deploy large-scale ML models. Both platforms have become significant open-source tools for AI and ML due to their ability to support the development and training of complex models on large datasets. They have been widely adopted by researchers and developers throughout the world and are regularly used in a variety of applications relevant to the medical technology industry. Significantly, they give smaller MedTechs a competitive advantage.    Disruptive effects of AI driven big data strategies The development and availability of big data and predictive AI help small medical technology companies enter markets, grow, and strengthen their competitive positions, which has the potential to change market dynamics. Over the past decade, several large medical technology companies have experienced their markets dented by small companies, which have successfully used open-source AI applications to leverage big data. For example, Philips Healthcare’s market was affected by the emergence of innovative offerings developed by new entrants using cloud computing services and big data from medical wearables. Above we described how Philips robustly responded to this and became a market leader in AI data-driven patient monitoring technology. Siemens Healthineers’ market share suffered from small MedTechs with innovative AI driven offerings. Further, the rise of digital imaging technology caused GE Healthcare’s market share to shrink. These vast companies have since developed AI driven big data strategies and bounced back. However, traditional MedTechs that fail to leverage big data and AI capabilities risk being left behind in an increasingly competitive digitalized industry.   Small MedTechs using big data and AI Examples of small MedTechs that leverage big data, AI, and ML techniques to capture share of large underserved fast-growing market segments include Brainomix, which was spun out of Oxford University, UK, in 2010 and serves the stroke market. Iradys, a French start-up specialising in interventional neuroradiology. Elucid, a Boston, US-based MedTech founded in 2013, which has developed innovative technology that supports the clinical adoption of coronary computed tomography angiography, and Orpyx Medical Technologies, a Canadian company that provides sensory insoles for people living with diabetes. These are just a few examples of small agile companies that collectively have helped to bifurcate and disrupt segments of the medical technology industry by developing offerings predicated upon big data, AI and ML that deliver faster, more accurate diagnoses to ensure that patients get the treatment they need, when they need it. Brainomex’s lead product offering is a CE-marked e-Stroke platform, which has been developed using data from images sourced across 27 countries including the UK, Germany, Spain, Italy, and the US and provides fast, effective and accurate analysis of brain scans that expedite treatment decisions for stroke patients. The platform has been adopted across multiple healthcare systems throughout the world, and for the past two years, England’s National Health Service (NHS) has been using the technology on suspected stroke patients. Early-stage analysis of the technology predicated on >110,000 patients suggests that eStroke can reduce the time between presenting with a stroke and treatment by ~1 hour and is associated with a tripling in the number of stroke patients recovering with no or only slight disability - defined as achieving functional independence - from 16% to 49%. With this disease, time is of the essence because after a stroke, each minute that passes without treatment leads to the death of ~2m neurons (nerve cells in the brain), which cause permanent damage. It can be challenging for health professionals to determine whether stroke patients need an operation or drugs, because the interpretation of brain scans is complicated and specialist doctors are required. Sajid Alam, stroke consultant at a large regional hospital in the UK, (Ipswich Hospital), reflected: “As a district general hospital, we don’t have ready access to dedicated neuroradiologists to interpret every stroke scan. Having Brainomix’s AI software gives us more confidence when interpreting each scan.” Intradys is a French start-up, which develops algorithms that combine ML and mixed reality to empower interventional neuroradiologists and help them enhance the care of stroke patients. Orpyx Medical Technologies provides sensory insoles for people living with diabetes who have developed peripheral neuropathy to help prevent foot ulcers. The insoles collect data on pressure, temperature, and steps and give feedback to the wearer and healthcare professionals. Elucid is a Boston-based MedTech founded in 2013. The company’s offerings are predicated on big data, AI, and ML to provide fast and precise treatments that improve the outcomes of patients with cardiovascular disease and reduce healthcare costs. Heart attack and stroke are primarily caused by unstable, non-obstructive plaque (the buildup of fats, cholesterol, and other substances in and on the artery walls) that often goes undiagnosed and untreated. Current non-invasive testing cannot visualize the biology deep inside artery walls where heart disease develops. Elucid’s lead offering is an FDA-Cleared and CE-marked non-invasive software to quantify atherosclerotic plaque.   Takeaways   The potential benefits for medical technology companies that leverage AI driven big data strategies include: (i) improved diagnoses and treatments, (ii) enhanced patient journeys and outcomes, (iii) cost savings, (iv) a better understanding of stakeholders’ needs, (v) superior decision-making, (vi) more effective products and services, and (vii) increased competitive advantage. Big data strategies may also be used to uncover insights from large datasets to develop predictive models that can automate repetitive tasks, optimize care processes, free up resources for healthcare professionals to focus on providing care, and staying ahead of the competition by providing greater insights into customer trends and needs. Medical technology companies that do not leverage AI driven big data strategies to develop innovative products for growth and competitive advantage potentially risk: (i) falling behind the competition in terms of product innovation, (ii) missing out on key market opportunities, as data-driven insights can help identify new trends and customer needs, (iii) struggling to keep up with the changing pace of technological change, as staying ahead of the competition requires a deep understanding of the latest developments in data-driven product development and (iv) losing the trust of customers, as they may be wary of MedTechs that do not use advanced technologies to develop their product offerings. Future significant growth for medical technology companies is more likely than not to come from AI driven big data strategies. Start collecting data.