From the January 2017 issue of HealthCare Business News magazine
By Jeroen Tas
Although artificial intelligence (AI) is high on the hype cycle, you are probably already reaping the benefits.
When you use Google Translate, talk to Amazon Alexa or see Facebook facial recognition, you are experiencing AI. The health care industry is one of the most interesting fields for AI, as there are vast amounts of data — most of which is unstructured — that can be transformed into relevant, timely and actionable information. These data sources include medical images, electronic medical records, lab results and genome sequences. But with the addition of information from wearables and sensors, the amount of available data is growing exponentially.
IDC estimates that by 2020 the digital universe — the data we create and copy annually — will reach 44 zettabytes, or 44 trillion gigabytes. The world of big data is expanding to the point that existing statistical methods and rule-based systems are insufficient to make sense of it all. We need AI to be able to make complex and unstructured data meaningful, to find the patterns that help us gain insights and predictive value.
While the methods date back to the ’70s, it’s only now that we have the affordable computing power, the storage and the data volumes to successfully leverage the technology. As the health industry moves to preventive models of health care, AI allows us to come up with better ways to diagnose and cure diseases, as well as lower costs and increase efficiencies in health care processes.
AI can help interpret and automate routine, time-consuming processes, help reduce errors, and provide clinical decision support and predictive analytics. One way this is being done is by using AI in programs that can automatically sift through patient data to present all relevant data in a “mission briefing” — a concept borrowed from gaming — providing relevant patient context for a radiologist.
Being able to record the radiologists’ preferences and adapt the user interface assists the clinician by offering tool sets and measurements driven by the understanding of the clinical context. These systems will remember how prior studies were interpreted, quantify information on the image and highlight progression over time. In diagnosing multiple sclerosis, which manifests itself as multiple dots on a medical image, subtle changes will be highlighted by the AI agent, which interprets the images and can see things that the human eye could easily miss.