By Alex Card
The human brain is a beautiful organ, capable of processing terabits of information at speeds many times faster than the world’s most advanced supercomputers—and with considerably less environmental impact.
But, as any brain owner will tell you, it’s not a perfect machine. Our grey matter encodes far more information than we can actually comprehend or reliably recall from long-term storage. We can only hold in our working memory about seven concepts at once, give or take two. For tasks requiring the analysis of massive datasets—like those produced by billions of healthcare encounters every year—computers offer a more useful sort of intelligence.
In the latest Definitive Healthcare special report, we asked 132 healthcare providers and executives about how their organizations use artificial intelligence (AI) and machine learning (ML) to improve care outcomes, develop more efficient workflows, and more. We also asked about those organizations’ plans for future AI/ML use, as well as the challenges they’ve faced implementing this technology.
Our survey revealed some interesting trends around AI/ML in the healthcare space—but before I dig into those, let’s pin down what intelligence means when computers are doing the “thinking”.
What is artificial intelligence?
Artificial intelligence is a technology that simulates natural intelligence, often for the purpose of learning, problem-solving, or decision-making. In healthcare, AI may perform tasks once handled by a provider or other healthcare professional, like analyzing clinical data, delivering insights to physicians, or controlling medical systems.
The term “AI” covers a broad swath of technologies, including augmented intelligence, robotic automation, natural language processing, and machine learning.
Machine learning is a type of AI that uses algorithms and data to replicate the way humans learn, incorporating new information to synthesize or improve a computer’s “understanding” of a concept. With the assistance of ML, healthcare professionals can more easily develop drugs, identify and treat disease, and manage administrative processes.
With the definitions out of the way, let’s explore the four big trends our report uncovered.
1. AI/ML is used most for oncology imaging and workflow improvement
AI is gaining ground, but most organizations are still primarily relying on good, old-fashioned human intelligence to get their work done. Just under one-third of survey respondents reported that their organizations currently use AI/ML, mostly in the Northeast and Southwest U.S.
Among those organizations employing AI/ML today, about 36% are using this technology for either computer-aided image detection for oncology or process/workflow improvement. Another 34% are using AI/ML to suggest more impactful care options.
While the human brain and eye make a highly effective team for recognizing patterns and anomalies in medical imaging, AI/ML-equipped computers can process millions of images in a fraction of the time. Combined with human intelligence, AI/ML provides an additional layer of accuracy and reliability that reduces misdiagnoses and ensures fewer concerns go unnoticed.
2. Healthcare professionals see major potential in process improvement and care suggestions
About 40% of healthcare professionals plan to incorporate AI/ML into their daily operations in the next two years.
Those who currently leverage this technology seem ready for more—nearly three-quarters have plans to expand AI/ML usage over the same period. Those who haven’t already bought in, however, are less convinced: only about 27% of non-users plan to adopt AI/ML tech in the near future.
Process/workflow improvement is the top area for intended AI/ML use, with more than two-thirds of respondents reporting interest in this area. No surprise there: AI can drastically cut the time necessary to create schedules, assign physicians to appropriate patients, measure staff productivity, and distribute clinical information. This puts more time back in providers’ hands, so they can focus more on patient experiences and outcomes—and avoid being overworked to the point of burnout.
More than half of survey respondents plan to implement AI/ML to improve care guideline consultation/suggestive care options. AI/ML systems can be trained to recognize patterns that human clinicians might miss, supporting the development of more impactful care plans and recommendations with less effort on the provider’s part.
3. Cost is keeping healthcare organizations from adopting AI
AI/ML technology simplifies all kinds of processes, but it’s far from simple to develop or maintain. For this reason, AI/ML can be fairly expensive to implement, especially on the scale of a hospital or a health system (where over a third of our respondents work).
Fittingly, about 57% of respondents said cost was their organization’s biggest hurdle to implementing and using AI.
While many AI/ML systems and tools are supported by their developers as part of a service arrangement, the guidance being provided isn’t always enough. Organizations commonly report challenges like lack of strategic direction, necessary IT infrastructure, or technical expertise. This points to a disconnect between the value of AI/ML technology and its understanding among those best suited to benefit it.
These aren’t the only issues, however. Respondents also expressed concerns about cybersecurity and AI, lack of regulatory guidelines from the FDA around AI/ML usage, and biases within AI’s detection ability.
4. Providers say AI’s biggest potential for patients is in faster, more accurate diagnoses
When it comes to patient care, about 40% of healthcare professionals believe AI will have the biggest impact on reducing the time of diagnosis and aiding in earlier disease detection. Nearly the same number of respondents see improved diagnosis accuracy as the greatest impact of AI/ML for patients.
About 38% of respondents believe AI-based improvements to workflow processes and existing technologies will make the biggest difference in patient care.
As indicated earlier, providers’ opinions on the impact of AI on patient care seem directly tied to whether they’re currently using AI/ML technology. Among all respondents, the average rating of AI’s value for patient care is 6.95 (out of a 10-point scale, with 10 representing “significant value”). For those actively using AI, the average rating is 8.17.
Don’t take our word for it—take AI/ML for a test drive
AI hasn’t won the hearts and minds of every provider, but it’s gaining ground. Most of those who currently use it are sold on its value—and they want more. Plus, many providers who haven’t yet taken the plunge are making plans to do so in the next couple of years.
If you want to dig deeper into the trends discussed here, check out the full Definitive Healthcare special report.
Ready to put this technology to work yourself? AI/ML technology is the heart—or brain, rather—of our healthcare commercial intelligence platform, supporting the delivery of clinical and market data, analytics and expertise that can help you make faster, more confident business decisions and improve patient outcomes.
In other words, we know how this stuff works and we’d love to show you. Sign up for a free trial today and see how AI-powered healthcare commercial intelligence can help you find success in a complex market.