As Devices Generate More Data, AI Is Becoming Indispensable for Medtech | Deloitte US has been saved
Limited functionality available
By Pedro Arboleda, managing director, Deloitte Monitor
Artificial intelligence (AI) is all around us. Even as I write this blog, the AI software in my computer is automatically correcting my misspellings, comma errors, and clunky grammar. It learns from my mistakes and gets better each time I write. As I head to the airport, my traffic app uses AI to continually recalibrate my optimal route, and my ride-sharing app uses the same technology to calculate my fare.
In the medical technology space, we have been discussing the potential of AI for a number of years. During AdvaMed’s annual conference in Boston last month, the buzz around AI was noticeably louder. There was an acknowledgement among many speakers and attendees that AI has become indispensable for medtech companies that want to pull clinical insights from the enormous pools of data being generated by their devices and other sources. Someone at this year’s conference described AI as a software engineer in a box. I thought that was an apt analogy. Like a software engineer, AI is always learning. But this engineer works around the clock with no breaks…and it teaches other engineers.
Is AI the future of medtech?
During the first half of 2019, investments in health-related AI topped $1.4 billion—up significantly from the first half of 2018.1 Some large medtech companies have made substantial investments in AI. Siemens, for example, has more than 40 AI-related patents, and more than 500 patents in the field of machine learning.2 Philips employs more than 500 data scientists and 60 percent of its research and development budget is now in software—some of which is destined for AI-specific applications.3
While AI is top of mind for many executives, few medtech companies have the internal resources needed to fully harness its potential. The volume of calculations needed for top-level research in AI has increased 300,000 times over the past six years, according to a recent report from the Allen Institute for Artificial Intelligence. Few data scientists have deep expertise in AI (just 5,400 worldwide according to Deloitte research), and they tend to be snatched up by deep-pocketed technology companies that are building self-driving cars or next-generation digital assistants. The starting salaries for an AI-savvy data scientist tops $1 million (not including signing bonuses or other benefits).4 Securing the talent and the resources needed to excel in AI is something that few medtech companies can do on their own.5
Partnerships could be a win-win
Medtech companies have historically focused on hardware-based scientific breakthroughs. Companies that don’t have the resources needed to build AI from the ground up might consider forming partnerships with technology companies that have already made those investments and are looking to create more use cases in health care. Some pharmaceutical manufacturers, for example, have forged relationships with technology companies that have AI that can help diagnose rare conditions and develop new therapies to treat them. Technology companies are also working closely with some health systems to analyze and learn from data on complex patient conditions.
These types of relationships could give health systems and life sciences companies access to the talent and computing power they need to generate clinical insights from the data they generate and manage. A meaningful give-and-take of expertise between medtech and consumer tech companies could be a win-win. Google’s DeepMind AI programs, for example, are being developed to read mammograms and head and neck CT scans.6 The programs are learning to spot signs of macular degeneration.7
While technology companies often have sophisticated AI capabilities, medtech companies have deep expertise in the clinical development of medical algorithms, such as translating data from an EKG lead into meaningful output that a physician can use. This clinical expertise and credibility with physicians could be useful to potential consumer tech partners. Moreover, consumer technology companies’ data science and AI expertise— combined with medtech’s ability to develop meaningful medical applications and algorithms—could lead to powerful offerings that will improve patient health.
Our recent report, Winning in the Future of Medtech, concludes that consumer technology companies and medtech companies could mutually benefit from greater collaboration on the use of AI. Some medtech companies have already joined forces with consumer technology companies to harness AI. For example, iRhythm Technologies, which manufactures wearable electrocardiograms, is working with Verily Life Sciences to develop ways to better screen, diagnose, and manage patients who are living with so-called “silent” atrial fibrillation.8
Standards, best practices, and guardrails are needed
Regulators are working to develop regulatory guardrails as AI applications take off in medtech. Earlier this month, the US Food and Drug Administration (FDA) released a draft framework detailing the types of AI/machine learning-based algorithm changes in medical devices that might be exempt from pre-market submission requirements.9 As part of the Consumer Technology Association’s AI initiative, AdvaMed, Google, Doctor On Demand and other organizations will work to develop standards and best practices for AI use cases in medicine and health.10
FDA recently approved a collection of AI algorithms embedded in a mobile X-ray device from GE Healthcare.11 The AI can help radiologists identify certain lung issues. So far, FDA has approved at least 30 AI algorithms—primarily diagnosis and monitoring applications—with radiology and medical imaging being the most pervasive application. Of the more than 100 medical imaging AI startups in 2018, the majority were for image analysis.12
A recent crowdsourcing simulation conducted by the Deloitte Center for Health Solution asked executives from digital health startups, medtech companies and technology companies which technology they thought had the most potential to transform the quality of health care. Nearly 80 percent of participants pointed to AI. AI—along with machine learning, reinforcement learning, and other subcomponents—can give medtech companies the ability to scrutinize enormous amounts of data, identify subtle patterns, and learn from the process.
During my flight today, the pilot will likely rely on an AI-enabled guidance system for most of the flight. Once we land, AI will help guide me to my hotel, and dinner choices—based on past choices—might pop up on my smartphone. AI is all around us today. In medtech, it is becoming indispensable.
1. Global Healthcare Report Q2 2019, (https://www.cbinsights.com/research/report/healthcare-trends-q2-2019/)
2. Siemens Healthineers press release, (https://www.siemens-healthineers.com/en-us/infrastructure-it/artificial-intelligence/our-expertise)
3. Philips lights on AI to target growth in healthcare, Financial Times, January 27, 2019
4. A.I. Researchers Are Making More Than $1 Million, Even at a Nonprofit, New York Times, April 19, 2018
5. At Techʼs Leading Edge, Worry About a Concentration of Power, New York Times, September 26, 2019
6. DeepMind’s health team makes the jump to Google with some NHS partnerships in tow, FierceBiotech, September 18, 2019
7. DeepMind has made a prototype product that can diagnose eye diseases, MIT Technology Review, April 1, 2019
8. iRhythm links with Verily to help diagnose ‘silent’ afib cases, Fierce Biotech, September 6, 2019
9. Clinical Decision Support Software; Draft Guidance for Industry and Food and Drug Administration Staff, Federal Register, September 27
10. Google, Fitbit, industry groups join to advance AI tools, Healthcare Dive, April 8, 2019
11. GE’s health unit wins first FDA clearance for A.I.-powered X-ray system, CNBC, September 12, 2019
12. AI just as effective as docs in medical imaging, study suggests, MedTechDive, September 25, 2019
Pedro Arboleda is a managing director in Deloitte Consulting LLP’s Strategy practice. With 18 years of experience in the Life Sciences industry, Pedro has focused his career on developing commercial and business development strategies for small-cap, mid-cap and multi-billion dollar medical technology companies. He has applied his knowledge of and passion for the convergence of health, predictive data analytics, cognitive technologies, behavioral science, and innovative solutions to transform business models and tackle challenging societal health issues for Deloitte’s medical technology clients.