A (friendly) ghost in the machine: Exploring AI in medical devices

Share this post

By Alex Card

Want to spook a professional writer? Tell them ChatGPT got a new update.

That’s a joke, of course. At Definitive Healthcare, we see the artificial intelligence (AI) and machine learning (ML) technologies behind OpenAI’s popular chatbot not as competition, but as powerful tools for human users—think force multipliers, not labor force replacements. Likewise, the World Economic Forum’s “The Future of Jobs Report 2020” predicted that while AI may displace 85 million jobs by 2025, AI could create 97 million new roles in that same period.

AI/ML technology may have some professional writers feeling a bit nervous, but the healthcare industry overall looks like it’s ready to embrace AI, however gradually. While concerns about learned bias and data security risks are reasonable, it’s hard to ignore benefits like increased workforce efficiency, improved diagnosis accuracy, and newly available time to spend on things to which humans are uniquely suited (like talking to patients).

Some of the most exciting advancements in healthcare AI are happening in the medical device space. The Food and Drug Administration (FDA) has approved at least 521 AI/ML-enabled medical devices since 1995, with more than half of those approvals occurring in the last three years. Doctors and patients around the country are benefiting from this technology every day, but there’s still plenty of room to grow.

So how are AI and ML being used in medical devices? Let’s take a look at the top use cases:

  • Disease diagnosis
  • Medical imaging
  • Remote patient monitoring

Accuracy matters for disease diagnosis

AI and ML are being used to develop medical devices and software that can diagnose diseases with greater accuracy and speed. These tools can analyze a patient's medical history, symptoms, and test results to identify potential health issues. AI/ML tech excels at identifying patterns in large data sets that may not be apparent to human doctors, making it ideal for population health management, too.

There’s hardly a specialty that hasn’t found a diagnostic use for AI/ML-enabled devices. In the last few years, the FDA has approved a urine test for kidney disease, multi-purpose ultrasound systems, a device that can diagnose autism spectrum disorder in children by tracking eye movements, a blood culture microbial test kit, and a software device that uses cameras to detect diabetic retinopathy—all equipped with AI/ML technology.

With so many applications, you might expect that AI has found its way into every doctor’s office and hospital exam room in the country. But not everyone’s quite on board yet.

Our August 2022 survey of 132 providers and hospitals executives found that only about one-third of organizations use AI/ML today. Still, respondents believed that improving the speed and accuracy of diagnosis were the two greatest patient-facing impacts of AI/ML.

AI makes quick work of medical imaging

Medical imaging assists in the diagnosis of countless diseases and conditions, and AI/ML’s superhuman speed gives it an edge in transforming these images into actionable medical intelligence. While a senior radiologist can take 20 minutes or more to assess a computerized tomography (CT) scan containing hundreds of images, an AI program can do the job in 20 seconds.

Over 75% of AI/ML-enabled medical devices approved by the FDA fall under radiology, so while your family practitioner may not be familiar with AI yet, your radiation tech almost certainly is. Applications range from computer-assisted tools for prostate cancer detection to software that makes bones easier to distinguish in MRIs to a device that generates contours of brain tumors to guide radiation treatment.

Several major industry players are pushing AI forward in medical imaging. In recent years, NVIDIA, the company that produces the graphics processing units underlying many AI/ML technologies, has worked with GE Healthcare and Nuance Healthcare Solutions to support and accelerate thousands of startups worldwide toward the development of AI to improve medical imaging accuracy.

Wearable AI keeps patients and providers informed

Remote patient monitoring (RPM) makes it easier for patients and their doctors to track and manage chronic disease and overall wellness without constant medical intervention. As with AI, this technology hasn’t achieved widespread adoption (check out my RPM report to see who the early adopters are) but it’s growing fast, and the two technologies are proving to be an effective pair.

AI/ML technology supports RPM by identifying emerging concerns or unusual readings, alerting users and doctors to significant changes, and transmitting relevant information to providers and payors for faster intervention and simpler recordkeeping.

One FDA-approved API integrates into existing electrocardiogram devices and transmits, analyzes, and annotates cardiac data for remote doctors. Another FDA-approved RPM system uses AI to collect health metrics from a variety of patient wearables to create data hubs for review by patients and their physicians. An especially smart smartwatch leverages AI to detect seizures—and alert caregivers—using accelerometer, temperature, and electrodermal activity data.

Amid the perpetual headache known as the healthcare staffing shortage, providers are always looking for opportunities to save time and maximize the efficiency of their efforts without compromising on the quality of care. AI/ML-enabled RPM devices can do just that—and even potentially save health systems and payors money by preventing emergency room trips and hospital readmissions.

Embrace efficiency with AI

Is this writer afraid of ChatGPT? So far, my colleagues and I agree that this AI tool is friend, not foe.

For everyone else seeking success in healthcare, AI and ML present some incredible opportunities for elevating the impact of their efforts, whether you’re caring for patients, running clinical trials, or trying to understand your market. 

Definitive Healthcare uses AI/ML technology to turn data, analytics, and expertise into healthcare commercial intelligence on providers, patients, facilities, and more. Want to see how it works? Sign up for a free trial today.

Home
  • Blog
  • A (friendly) ghost in the machine: Exploring AI in medical devices