Busy clinicians barely have enough time to eat and sleep, let alone keep up with all the changes around them. And the changes are coming fast. Blink once and the entire world is talking about artificial intelligence (AI). Blink again and they’re talking about AI in healthcare. Part of our role at Medmastery is to help you—the busy clinician—stay on top of your game. With that in mind, here are seven examples of AI in healthcare that you should know.
What is AI?
Before jumping in, it’s worth reviewing some basic terms and concepts that you may come across:
- AI is a very broad term that refers to machines imitating intelligent human behavior.
- The application of AI as we know it is thus far limited to specific tasks (like a car driving itself), a concept known as narrow AI or weak AI.
- The idea of artificial general intelligence (AGI)—where an AI system would have full human cognitive ability and perform unfamiliar and unrelated tasks—is known as strong AI and is probably far off.
- Machine learning is a subset of AI that allows machines (computers) the ability to learn without being explicitly programmed. Because most advances in AI have involved machine learning, the terms are often used interchangeably.
- The general idea of machine learning involves giving the machine a lot of data along with a task (a goal). Eventually, the machine can accomplish the desired task. Of course, the technical side of the learning can be quite complicated.
- Deep learning is a type of machine learning that uses artificial neural networks, modeled loosely after the way a human brain works. The tasks accomplished can be more complex.
What are some examples of AI in healthcare?
Ample data is key for successful machine learning. With the explosion of data in healthcare, it’s not surprising that AI is making significant inroads into the field. The uses—and potential uses—of the technology are limited only by the imagination. These examples are just the tip of the iceberg, chosen to help you get a general idea of how AI is being used to make positive change.
A well-trained AI can be used to identify a specific type of abnormality, significantly increasing the number of images that can be assessed for that abnormality in a given amount of time. AI is also being assessed in the next generation of radiology tools that could eventually obviate the need for obtaining tissue samples to make certain diagnoses.
AI-assisted robotic surgery
Robots can serve as highly effective assistants for various types of procedures. Given the ability to provide magnified and three-dimensional views not available to the human eye—along with the precision control of robotic arms—robot-assisted surgeries can be associated with less complications and speedier recovery times. As of now, humans have been driving the robots, though self-driving robots will have a role in performing some tasks, taking the burden off the surgeon.
The example that is often given is the successful use of an AI algorithm (involving deep learning) in the diagnosis of skin cancer. A similar concept can be applied to a variety of conditions, helping to prioritize those patients who need timely care.
Virtual nursing assistants
Available 24 / 7, such assistants can provide quick answers to simple patient questions, particularly in between office visits. These assistants can also be used to help triage patients to the appropriate care setting.
Connected medical devices
There has been a proliferation of connected medical devices in recent times, adding to the growing heap of available data. The next step is to have such devices process the data and recommend or administer treatment. An example would be the automated optimization of insulin pump settings using the data from multiple blood glucose readings.
Prescription error recognition
Given AI’s strength in processing data, it makes sense to use the technology to prevent prescription errors. Examples of things that could get flagged are prescriptions that don’t fit the documented past medical history and medication doses that don’t match weight, body surface area, or kidney function.
There’s no need to discuss the obvious administrative burden in delivering healthcare. AI (specifically natural language processing) is already being used in voice-to-text transcription for note writing. Similar technology could be used to order tests and write prescriptions, simplifying the time consuming tasks that define day-to-day patient care.
Of course, the point of AI in healthcare should be to help healthcare providers and improve patient care. Getting there will involve some growing pains, including steep learning curves and the potential overuse of unnecessary technology. But those barriers shouldn’t—and won’t—halt the inevitable integration of AI into the world of clinical medicine.