The AI revolution in healthcare is upon us. Institutions like The Mayo Clinic, Memorial Sloan Kettering Cancer Center, Massachusetts General Hospital, and the United Kingdom’s National Health Service have already begun using AI in various capacities, and it seems there’s a new study every day demonstrating AI’s incredible power to crunch unimaginable amounts of data, generate new insights, or diagnose diseases. Not to mention everything happening outside the healthcare world with virtual assistants, autonomous vehicles, fraud detection, and natural language generation—just to name a few. But what exactly is AI? And how are we really using it?
What Is AI?
AI studies the design of an intelligent agent—a system that acts accordingly in relation to its circumstances and goals, learns from experience, and makes appropriate choices given its limitations. There are many approaches to AI: cognitive simulation, logic programming, embodied intelligence, and artificial neural networks, among others. However, it seems that machine learning has demonstrated the most promise in healthcare applications.
Machine learning refers to the study of the algorithms and statistical models’ computers use to execute a given task without explicit instructions and instead rely on patterns and inference. Machine learning algorithms use sample data to build mathematical models that make predictions or decisions without being explicitly programmed to do so. There are several different types of machine learning algorithms. Let’s see what machine learning can do for physicians.
AI is already showing tremendous promise in disease diagnosis. In a study1 from Annals of Oncology release last year, researchers in Germany, the US, and France trained a convolutional neural network (CNN) (a type of deep learning algorithm) to identify skin cancer by showing it more than 100,000 images of malignant melanomas and benign moles. They then tested the CNN against 58 international dermatologists and found that the CNN missed fewer melanomas and misdiagnosed benign moles less often than the group of dermatologists. A recent study2 from The Lancet Digital Health found that “the diagnostic performance of deep learning models to be equivalent to that of health-care professionals.” Additionally, a study3 from Nature Medicine found that an AI trained to identify pediatric diseases could diagnose glandular fever, roseola, influenza, chicken pox, and hand-foot-mouth disease with 90-97% accuracy.
These studies are obviously limited in scope and preliminary, but they do point to a bright future in disease diagnosis. The key to their power is the ability to sort through and manage massive amounts of data. A machine learning algorithm can process more images in less time than a human and as these algorithms get more sophisticated, we can expect they’ll be better at diagnosing rare diseases or unique presentations of common ones than human healthcare providers and will increase overall efficiency in healthcare.
Beyond diagnosis, AI is also giving physicians the power to design treatment plans. At the University of Toronto’s Department of Mechanical & Industrial Engineering, researchers have developed AI software to mine radiation therapy data and then apply it to an optimization engine to develop treatment plans. They applied this tool to a study4 of 217 patients with throat cancer and found that the AI-generated therapies achieved comparable results to patients’ conventionally planned treatments and in less time (20 minutes). Additionally, researchers at MIT have developed an AI model5 which learns from patient data to adjust doses and find an optimal treatment plan for patients suffering from glioblastoma, the most aggressive form of brain cancer.
The examples I’ve just listed are but a small sample of everything happening in AI in healthcare and they’re an even smaller sample of what is to come. In the coming years, AI is expected to help reduce healthcare costs and improve patient care by streamlining workflows and making hospitals more efficient and help identify individuals at risk of readmittance. Additionally, some pharma companies are already using AI in the drug development process and the proliferation of wearables and other medical devices is continually helping physicians detect diseases earlier and treat them more effectively. The AI revolution is here and I’m excited to see what the future holds.
About the author
Dionne has more than 18 years of experience in the healthcare industry, 16 of which she spent leading market research, analytics and marketing innovation at large pharma companies.
There’s a moment in “The Wizard of Oz,” after Auntie Em and Uncle Henry’s house falls to the ground, where Dorothy gets up off the bed and opens the door which takes her into a Technicolor world. Medical education and healthcare marketing are...
“A display connected to a digital computer gives us a chance to gain familiarity with concepts not realizable in the physical world. It is a looking glass into a mathematical wonderland1.” -Ivan B. Sutherland, “Father of Computer Graphics” and...
Experiential learning allows for the effective communication and retention of complex medical education content. Techniques like interactive video, live polling, gamification, and mixed reality are helping the message stick in the minds of our...