Artificial Intelligence (AI), machine learning, and deep learning are taking the healthcare industry by storm. They are not pie in the sky technologies any longer; they are practical tools that can help companies optimize their service provision, improve the standard of care, generate more revenue, and decrease risk. Nearly all major companies in the healthcare space have already begun to use the technology in practice; here I present some of the important highlights of the implementation, and what they mean for other companies in healthcare.

General gains for the industry

The best metaphor I found describing the importance of AI is presented by Bertalan Meskó in one of his articles. It seems that the question is not “if” but “when” AI will revolutionize the healthcare.

It took some time for the medical community to accept the stethoscope. It will also take a while to recognize A.I. as a full-fledged health tool – despite its vast potential to revolutionize healthcare. Yet, it is so powerful that when it will finally take its rightful place in healthcare, it will displace the stethoscope as its symbol.

Bertalan Meskó, MD, PhD

AI, machine learning, and deep learning are already increasing profits in the healthcare industry. For example, according to research firm Frost & Sullivan by 2021, AI systems will generate $6.7 billion in global healthcare industry revenue. In 2014, they only generated $634 million—that’s a 40 percent compound annual growth rate.

Investment into these technologies is booming. In 2014, the cross-industry average revenues’ spending on IT was 3.3, but for healthcare providers, the average spent was 4.2 percent. Around 35 percent of healthcare organizations will implement artificial intelligence solutions within the next two years — and over half of them plan to follow suit within the next five years. I’ve analyzed the trend and recognized that an increase has been driven by a variety of needs specific to running any business in healthcare, including the need to create electronic healthcare records that are secure enough to comply with privacy laws. Between 2012 and 2017, penetration of electronic health care records grew from 40 to 67. I find this statistic crucial for the future of machine learning in healthcare because the availability of data is necessary for the further advancement in this topic.
At the same time, healthcare AI deals also grew in significant numbers, expanding from 20 such deals in 2012 to almost 70 by mid-2016. Venture capital investment in AI-driven medical technologies has also exploded, an open recognition of the area as promising one that is likely to deliver, at least in the eyes of tech-savvy money people. In 2012 AI-driven healthcare projects such as robotics, machine learning (ML), and computer vision totaled $30 million; in 2016, that area of investment topped $892 million.

According to a report from Accenture, the greatest near-term value for healthcare businesses right now exists in these top three applications:

  • robot-assisted surgery ($40 billion)
  • virtual nursing assistants ($20 billion)
  • administrative workflow assistance ($18 billion)

However, other areas I examined, such as medical imaging are also very promising—particularly because they are meeting such tremendous needs. For example, right now more than half of the world’s population has no access to medical imaging because those technologies that are not AI-assisted are expensive, unwieldy, and demand impractical levels of training. In total, Accenture estimates that AI will be able to address at least 20 percent of unmet clinician demand by 2026.

The takeaway: more money is going into practical AI-driven applications in the healthcare industry because those applications are generating revenue. And that’s is just a beginning.

Specific applications

To get a better sense of how AI and machine learning are transforming the healthcare industry now, it’s useful to consider specific cases. That is why I have gathered some of the more fascinating applications of the technologies in healthcare right now, which also demonstrate the practical value of these cutting-edge technologies.

Identifying tuberculosis in the developing world

Identifying patterns in images is in my opinion one of the strongest points of existing AI systems, and researchers are now training AI to review chest X-rays and identify tuberculosis. This technology could bring effective screening and evaluation to TB-prevalent regions that lack radiologists.

AI for treating war veterans with post-traumatic stress disorder (PTSD)

The Tiatros Post Traumatic Growth for Veterans program partnered with IBM Watson to use AI and analytics to ensure more veterans with PTSD would complete psychotherapy. Using these technologies, they achieved a 73 percent completion rate, up from less than 10 percent. As many as 80 percent of veterans with PTSD who finish a treatment program within a year of diagnosis can recover, according to statistics from the Department of Veterans Affairs. Approximately one in five of the 3 million veterans of Afghanistan and Iraq wars suffer from PTSD.

Detecting brain bleeds

Israeli healthcare tech company MedyMatch and IBM Watson Health are using AI to help doctors in hospital emergency rooms treat stroke and head trauma patients more effectively by detecting intracranial bleeding. The AI systems use clinical insights, deep learning, patient data, and machine vision to automatically flag potential cerebral bleeds for physician review.

Optimizing administrative workflow and eliminating waiting time

Administrative and assistant work is a prime area for AI. According to Accenture, timesaving workflow capabilities such as voice-to-text transcription have the potential to eliminate tasks like ordering tests and prescriptions and writing notes in charts for medical professionals—anything that concerns non-patient care. This amounts to a savings of 17 percent of doctor work time and a whopping 51 percent of registered nurse work time.

AI could also prioritize doctor emails and assist patients in resolving simple medical issues without the help of doctors, optimizing schedules from both sides. For example, the startup Scanadu’s natural language processing program allows patients to get their lab results explained to them by an app, saving both patient and doctor time and money. Nuance Communications has unveiled a similar product in the form of a virtual assistant that can explain test results and deal with basic patient concerns. The healthcare organizations that implement these technologies first will see the most benefit because they will have the most time to build knowledge libraries.

Detecting Alzheimer’s disease

It now takes AI-enabled robots less than one minute to diagnose Alzheimer’s disease with about 82 percent accuracy based on speech patterns and voice—and that level of accuracy is only growing. The AI systems can attend to the length of pauses between words, any preference for pronouns over proper nouns, overly simplistic descriptions, and variations in speech frequency and amplitude. While all of these factors are very tough for human listeners to note and detect with high levels of accuracy, AI systems are objective and quantifiable in their analysis.

Cancer diagnosing

Traditional methods for detecting and diagnosing cancers include computed tomography (CT), magnetic resonance imaging (MRI), ultrasonography, and X-ray. Unfortunately, many cancers cannot be diagnosed accurately enough to reliably save lives with these techniques. Analysis of microarray gene profiles is an alternative but relies on many hours of computation—unless that analysis is AI-enabled. Stanford’s AI-enabled diagnostic algorithm has now been proven just as effective at detecting potential skin cancers from images as a team of 21 board-certified dermatologists. Startup Enlitic is employing deep learning to detect lung cancer nodules in CT images—and their algorithm is 50 percent more accurate than an expert thoracic radiologists working as a team.

Other healthcare companies are going past diagnosis and on to treatment and even cures with the help of AI. Insilico Medicine is finding new drugs and treatments with deep learning algorithms, including new immunotherapies. These gene therapies use the cells of each individual patient to model their own biology and immune systems.

AI makes these cures work because it can design combination therapies and identify incredibly complex biomarkers by performing millions of experiments in simulated form at lightning speed.

Robo-assisted surgery

When it comes to value potential, robot-assisted surgery is at the head of the AI-enabled class. AI-enabled robotics can enhance and guide the precision of the surgical instrument by integrating real-time operating metrics, data from actual surgical experiences, and information from pre-op medical records. In fact, Accenture reports that these advances made possible by AI-enabled robotics include a length of stay reduced by 21 percent.

Studying various solutions, I find Mazor Robotics most promising. It is using AI to minimize the invasiveness and maximize the customization of surgical operations on areas with complex anatomy—such as the spine. The AI system helps the surgeon plan where implants will be placed using CT scans before the patient is present, and Mazor’s robot arm for spinal surgery guides the movement of the surgical instruments, ensuring a high degree of precision.

Worse yet, it gets smarter!

We shouldn’t get too excited about how Machine Learning and AI are going to change the healthcare as we know it. There is a lot of work to be done by people if we want to see those changes happening. In fact, the human factor is crucial for success in this case. Moreover, pumping up the topic may create unnecessary pressure on parties involved in the process. Great reality check is provided by Kevin Pho, MD in his article which shows the down-to-earth attitude that is, in the end, beneficial for the industry development. According to Kevin, we need to shift our focus to the translation of our priorities and workflows to models so that technology will be able to improve healthcare in the way we need it to.

Technology is great. But people and process improve care. The best predictions are merely suggestions until they’re put into action. In healthcare, that’s the hard part. Success requires talking to people and spending time learning context and workflows — no matter how badly vendors or investors would like to believe otherwise. It would be fantastic if health care could be transformed by installing software that assumed your workflows and priorities.

Kevin Pho, MD

The bottom line

There have always been cases of over-hyped technologies throughout history, but AI and machine learning are absolutely not among of them.

Even as we stand at the nascent edge of the technology, with its potential only barely understood, the healthcare industry is experiencing an influx of productivity and revenue thanks to AI and machine learning.

Most major healthcare players are already investing in AI, recognizing major role of this technology in the future of the industry. Where it takes us from here will be exciting to see, but a well-informed, studied opinion with the right knowledge will have the best chance of predicting its path.