At Google’s I/O developer’s conference in May, Google CEO, Sundar Pichai blew minds by demonstrating Google Duplex, a feature of Google Home and Assistant, which made a simple phone call to book a hair appointment.
If you missed the demonstration, it was an ordinary call made extraordinary because the conversation between Google’s machine and the salon’s receptionist was indistinguishable from one between two humans.
The demo ended up being an incredible, viral moment that highlighted the power of modern Artificial Intelligence for a wider audience. It also helped showcase how we’re only just beginning to glimpse the potential of AI, and there are still plenty of concerns around its abilities.
Forward-thinking minds like Stephen Hawking and Elon Musk have all warned about the consequences of AI, and it’s worth wondering about its imminent application in an industry as crucial to human survival as health care.
As venture capital firm Rock Health notes, health companies are leveraging AI and machine learning and raising a ton of money in the process — $2.7 billion from 2011 through 2017, to be exact.
A lot of the enthusiasm for the burgeoning technology comes from the belief that it has the power to revolutionize a wide range of areas within the industry, from creating cutting-edge medical devices to reducing misdiagnosis, advancing precision medicine to delivering faster, better care to at-risk patient groups.
That’s not to say everyone is jumping on the AI bandwagon blindly. Many are also weighing issues like patient perception, privacy concerns and potential disruption. Plus, the medical industry has seen its share of new, heavily-touted, messianic tech that hasn’t panned out. To gauge the debate, we put together some current pros and cons of artificial intelligence in healthcare.
Pro: Improving Diagnosis
Studies on diagnostic errors in the U.S. report overall misdiagnosis rates range from 5 percent to 15 percent and, for certain diseases, are as high as 97 percent.
Misdiagnosis is an understandable problem for doctors, as the World Health Organization’s International Statistical Classification of Diseases and Related Health Problems (ICD) lists about 70,000 diseases in total, with fewer than 200 presenting actual symptoms.
Assuming it’s been loaded with all the relevant data, an AI-equipped product has the potential to sift through disease data, clinical studies, medical records, genetic information and even a patient’s health records far quicker and more efficiently than a human physician for a more accurate diagnosis.
Con: AI Training Complications
According to Dr. Robert Mittendorff of Northwest Venture Partners, one significant challenge to AI in health care is the lack of curated data sets, which helps in training the technology to perform as requested through surprised learning.
“Curated data sets that are robust and have both the breadth and depth for training in a particular application are essential, but frequently hard to access due to privacy concerns, record identification concerns, and HIPAA,” Mittendorff as says in a recent Topbots article.
Pro: Better Serving Rural Communities
AI could benefit patients living in rural communities, where access to doctors and specialists can be tough. According to Stanford Medicine data, fewer than 10 percent of physicians practice in these communities.
At the Association of Academic Health Center’s 2017 Global Issues Forum, Dr. Yentram Huyen, General Manager, Genomics & Data Exchange, Health & Life Sciences, at Intel said that one way to address that problem is through collaboration for better data.
“Health centers should collaborate on the data, enabling an idea of federated data analytics,” Huyen says, according to Elsevier.com. “It is critical to break down the information silos. We have to think about how we’re going to collaborate and share the data to form [health care] partnerships.”
Con: Change is Tough
The health care community is still somewhat jaded by the last technology that was going to revolutionize the industry, electronic medical records (EMR).
EMRs were supposed to make everyone’s job easier, from the billings clerk all the way to the physicians. For all its benefits though, many found implementation to be a costly and time-consuming disruption to practices. And, as anyone who has experienced a new technology rollout at a company can attest, if things aren’t handled correctly, widespread adoption can be a major issue.
So for AI to be accepted by the medical community at large, it’s going to require, not just proof that it works, but a project plan that includes input from all stakeholders and evidence it’s worth the investment.
As Google’s demonstration showed the world, AI will be capable of handling complex and unexpected questions as long as it has plenty of good data to begin the process of deep learning.
The people serving in health care and those who supply goods and services to the market would be smart to develop and share a mutual understanding of AI. One thing everyone seems to agree on is it’s just a matter of time before we see it implemented in our health care system.