AI-Human Hive Mind: A new Diagnostic Tool for Doctors?
The Future of Healthcare: Part 3
Having made accurate predictions about the outcome of the Oscars, presidential elections, Kentucky Derby and World Series, tech company Unanimous AI now also aims to improve medical diagnostic accuracy. The first study was a success and may prove that humans and AI are better when they team up…
The blog series Future of Healthcare is aimed at healthcare professionals and discusses the most striking breakthroughs and developments in the healthcare sector.
In a small study conducted by Stanford University, eight radiologists were asked to diagnose pneumonia from chest X-rays. Diagnosing pneumonia on X-rays is quite difficult, as the disease shows many similarities to various other diseases. But with the help of AI-powered technology, the radiologists showed that they were more successful at diagnosing pneumonia than individual doctors (33% more accurate) and 22% more accurate than CheXNet, a machine-learning program created by Stanford University. The latter outperformed radiologists at diagnosing pneumonia from X-rays last year.
What is AI-Human Hive Mind?
The idea behind the AI-Human Hive Mind is that artificial intelligence amplifies the intelligence of human groups by using algorithms that are modeled after natural swarm behavior. Many species, like birds, fish, bees and ants amplify their group intelligence by forming swarms, schools, flocks and colonies. Working together as a unified system gives such formations the ability to outperform individual members when making decisions and solving problems. This phenomenon is called “Swarm Intelligence”, an ability that humans did not naturally evolve, as they are not able to create tight feedback-loops amongst their members, which plays a key role in swarm intelligence.
However, with the use of high-speed networking technology, humans from all over the world can now also connect with each other. The only thing needed, is the precise technology to turn the connections into a real-time system with closed feedback-loops among the members. This is where Unanimous AI comes in. The tech company created Swarm AI®technology, which connects artificial intelligence with human minds:
“It provides the interfaces and AI algorithms to enable“human swarms”to converge online, combining the knowledge, wisdom, insights, and intuitions of diverse groups into a single emergent intelligence. From predicting sports and financial trends, to assessing the effectiveness of advertisements and movie trailers, real-time swarms have been shown to significantly amplify intelligence.”
Application of the Swarm AI-system in the pneumonia study
The Swarm AI-system is not to be confused with crowdsourcing, in which participants simply respond with yes or no. Instead, each radiologist controlled a small icon on the screen that enabled to influence the group consensus by pushing it toward their own opinion. All the X-rays were analyzed in real-time with the other radiologists contributing their opinions simultaneously. As the doctors contributed their opinions, the AI algorithms monitored their behavior, by looking at the movement of their icons over time.
To achieve the best outcome, it is not enough to just know what the opinions of the doctors are. What is really important, is to know the different levels of confidence. For example, some doctors took longer to come make a diagnosis, while others changed opinion several times. Then the preferences were combined into a specific choice.
Benefits of Swarm AI
First of all, swarm AI technology enables each doctor to influence the diagnosis on an equal base. During meetings in large hospitals to discuss difficult cases, it may happen that the more introvert doctor has less influence on the outcome of the meeting. Swarm AI enables multiple doctors to agree on a case, by taking out the personalities and politics of the decision-making process. Secondly, human knowledge and experience is put to use in the most effective way. As CEO and founder of Unanimous AI, Louis Rosenberg, put it:
“We shouldn’t throw away human knowledge, wisdom, and experience. Instead, let’s look at how we can use AI to leverage those things.”
What are the future medical applications of Swarm AI?
Although it is not likely that Swarm AI technology will be used by radiologists on a daily base, it may be a very valuable tool in specific cases, where international experts are asked for their opinion. However, Swarm AI may prove it’s worth as a diagnostic tool for many other medical conditions. The Stanford research team plans to conduct a larger study at the Stanford University Medical Center, in which they use actual patient cases.
The following video shows how Swarm AI works:
