Doctors Vs. Data: The New Frontier of Machine Learning Driven Medicine

Google+ Pinterest LinkedIn Tumblr +

From robot-assisted surgeries, to automated anesthesiologists, to machine-learning based diagnoses, the medical profession is on the cusp of a technological, data-driven revolution: a new frontier of medical automation.

Despite technical advances, though, numerous barriers to medical automation adoption have cropped up: prohibitive costs, liability issues, and even pushback from doctors who do not trust machines to do their jobs. Because of these barriers, patients are receiving subpar treatment, are paying more for very basic procedures, and spend more time in the hospital receiving treatment.

We need to face the truth: software and robots are infinitely better at performing certain vital medical tasks. They can learn from a corpus of data and apply diagnoses based on precedent better than a human working from experience or memory ever could. They can perform surgery with more precision than a shaky human hand could. And they can even administer anesthesia better than a human anesthesiologist — a position that is currently extremely well paying for the level of education and expertise that it requires.

It is therefore inevitable that the medical profession will become increasingly automated — the question is how strong the pushback will be.

NewtonX assembled a large panel of senior healthcare professionals including 50 senior practitioners, 50 SVP level executives at large Healthcare providers, and 50 VP-level executives at large Healthcare providers. This panel has been leveraged both for qualitative as well as quantitative insights, both of which inform the angles in this article.

Data-Empowered Robots Take Over Procedures

Even the most sophisticated AI-based machines cannot do the work of most doctors. But it is undeniable there are key applications for robotics that enable better treatment options and lower risk for patients.

According to the NewtonX panel of 50 Healthcare Providers, 22% of the work currently done by medical personnel could be automated either through the use of simple Digital Solutions, more advanced AI or Automation, or even Robotics. However, there is significant variance across specialties, as well as in the kind of work required (e.g., complex diagnostics are not considered automatable by our experts, while 71% of all administratie work is considered automatable).

22% of the work currently done by medical personnel could be automated. Click To Tweet

1. Robotic Assisted Surgery

Robot-assisted surgery became a viable option in 2000, when the Da Vinci Surgical System — a minimally invasive robotic surgeon that is capable of performing complex surgeries — was approved by the FDA. Since then, over 1.75 million robotic surgery procedures have been performed, with “better visualization, increased precision, and enhanced dexterity compared to laparoscopy” according to the NIH. Recently, machine learning has helped increase the precision and efficacy of these robots by enabling real-time optimization of movements based on large data sets of past surgeries, body images, vitals, and the live information coming from sensors. A veteran expert from NewtonX (Head Surgeon in a large European Hospital) declares: “Surgery is right now at the crossroad of Big Data and the Internet of Things. The impact on procedures is already huge and will not stop growing.” The estimated number of adverse effects per procedure is less than .6%.

That said, while many surgeons welcome the use of the Da Vinci (or one of the few other systems such as MAKO Surgical’s RIO Robotic Arm Interactive System for orthopedic surgery), these robotic forms of medical automation are extremely expensive. The Da Vinci system average cost is between $1.5M and $2M according to our healthcare panel, which makes quite unaffordable for small and medium sized hospitals. To boot, it also comes with a steep learning curve: for robot assisted upper tract surgery, studies recommend between 30 and 40 procedures for training purposes. Considering the cost of human surgeons, this learning curve can also be prohibitively expensive for all but the largest hospitals.

2. Computer Vision-Based, AI-enabled Diagnosis

Diagnoses are made through simple pattern recognition. For a human, this requires years of advanced training and experience. For a machine, this requires a robust set of training data.

Google’s DeepMind is currently developing an AI-based machine to detect breast cancer more accurately than human pathologists looking at a Mammogram can. Machine learning algorithms have been able to detect the presence or absence of TB in X-rays with 96 percent accuracy — higher than any human radiologist. A startup called Optellum is working to commercialize an AI system that diagnoses lung cancer by analyzing clumps of cells found in scans — and they claim their system could cut costs by $13.5B. While many of these systems have not yet been deployed for widespread use, they’ve created major buzz in the medical field, and numerous studies have predicted that they will see mainstream adoption in the coming years.

3. Data-driven Anesthesiology

The pushback against robotic anesthesiology is one of the most interesting developments to have occurred in medical automation. In 2015, Johnson & Johnson released a machine called Sedasys that was approved by the FDA to administer the sedative propofol. After years of aggressive pushback from The American Society of Anesthesiologists, the machine’s uses were limited to colonoscopies and endoscopies. Even still, the benefits of Sedasys were immense: anesthesia administered by an anesthesiologist for a colonoscopy can cost between $1,000-$2,000; by contrast, Sedasys cost only $150-$200 per use.

And yet, the machine was pulled off the market after only a year due to poor sales. According to most news reports of the incident this was due in large part to aggressive pushback from anesthesiologists themselves, who warned that no machine could replace them. Sedasys was cost-effective, precise, and shortened hospital stays — but human distrust ended its potential.

“It’s not replacement; It’s Displacement”

While it’s understandable that doctors are concerned about medical automation — especially considering the heavy loan burdens that many doctors carry — the reality is that machines will not replace doctors; they will just displace them.

“Automation, AI and Robotics are not here to take jobs from Doctors or Medical Personnel,” said a C-level Executive at a large chain of clinics in Texas. “They are here to take the pain or repetition out of the work and help medical professionals focus on the value-add activities. We should not talk about replacement, we should talk about displacement”

Radiologists may not need to read X-rays to determine whether a patient has breast cancer anymore; but they will need to discuss treatment options, answer patient questions, and lend the human aspect of their expertise to patients. Furthermore, doctors will always be necessary for verifying data and feeding robots new data about emerging diseases and information about how diseases spread.

That said, according to our panel, big data and machine learning in pharma and medicine will generate a value of up to $100B annually, based on better decision-making, optimized innovation, improved efficiency of research/clinical trials, and new tool creation for physicians, consumers, insurers, and regulators. There is a massive opportunity right now for investing in productivity and care enhancing types of automation — and eventually, most healthcare organizations will get onboard just to compete.

Big data and machine learning in pharma and medicine will generate a value of up to $100B annually. Click To Tweet

This increase in automated technologies may result in layoffs, but may also simply result in better processes (not reading magazines in a waiting room for two hours every time you go to the doctor). And if hospitals and other providers can increase efficiency to the point where fewer doctors are needed to treat the same number of patients, this could have positive ramifications on healthcare costs for consumers.

There’s One Thing No Machine Can Do Better Than a Doctor

Machines can only learn from precedent; they cannot ideate new ways of diagnosing, they cannot identify new diseases, and they cannot hypothesize new treatment methods. Because of this, the role of the doctor in our society will always be privileged, and will never disappear. Investigation, the willingness to challenge, ideation for better ways of doing things — these are all uniquely human characteristics.

65% of medical professionals believe increased technology adoption will result in better patient outcome. Click To Tweet

In fact, Alphabet, Google’s parent company, recently demonstrated just this through its health subsidiary, Verily. Last week, the company announced that it had developed an AI eye scan that can predict cardiovascular risk as accurately as a blood test can. While the machine may be doing the actual prediction, it was human creativity that is responsible for testing the ability of an eye scan to do the same thing that a blood test could do.

What The Immediate Future of HCPs Looks Like

Only 15% of the experts in our Senior Insurance Executives panel see cost benefits coming out of new technology in the next five years. However, 65% believe increased technology adoption will result in better patient outcome.

According to the Chief Medical Officer at a large Hospital in the Seattle area, “Surprisingly, the most immediate effects of investments in technology for patients are around the experience and not the actual medical outcome: reduction of administrative friction, increased speed of admission, or speed of procedure. However, the improved medical outcomes are also there, it is just more difficult for patients to see them given they only see their case and not the broader picture across multiple patients”

The role of the doctor may change to be more research focused on one side, and more patient-facing on the other, but no matter how advanced medical automation becomes, it will never replace human creativity.


The data and insights in this article are sourced from NewtonX experts. For the purposes of this blog, we keep our experts anonymous and ensure that no confidential data or information has been disclosed. Experts are a mix of industry consultants and previous employees of the company(s) referenced.


About Author

Germain Chastel is the CEO and Founder of NewtonX.

Comments are closed.