In a recent post on the Augmedix blog, Joe Marks, company board member and executive director of the Center for Machine Learning & Health at Carnegie Mellon University, argues that given the current state of computer technology, IA—intelligence amplification—is a better choice than AI alone when it comes to medical note taking.
The best-known example of AI and machine learning came into being in 1987, when Professor George Kondraske developed a “character pattern recognition and communications apparatus.” We know this better as predictive text and autocorrect. The computer guesses what it thinks you intend or intended to type based on previous text-message examples and suggests the best guess to you as a completion or correction. The more you use predictive text and autocorrect over time, the better the program will become at selecting the right words.
The same algorithmic concepts behind predictive text and autocorrect can, in principle, be used for applications such as medical note taking. But while autocorrect errors might be embarrassing or funny, errors made by AI/ML algorithms in medical note taking can be deadly.
The goal of Augmedix is to populate electronic health record documentation from just the spoken interactions between care providers and patients. To do this, there are three subtasks an AI has to perform:
- Speech recognition—turning the spoken words into text words.
- Named entity and relationship recognition—recognizing medically meaningful phrases in the stream of text words.
- Interpretation—figuring out what the recognized phrases actually mean, so that accurate medical information can be recorded in the EHR document. This is where current AI/ML technology falls very short.
One example Marks provided of AI going drastically wrong is when it interprets the phrase “I’m doing well, can’t complain.” to “Change in bowel habit, do not resuscitate.”
“As we can see,” Marks wrote, “fully automated AI systems currently fall apart when interpreting more complex sentences that involve significant context.” And until that changes, Marks argues that the approach to medical note taking should not be AI, but IA—intelligence amplification.
IA refers to the effective use of AI methods for augmenting human intelligence, and this is exactly the technology Augmedix uses. Remote document specialists are helped in their note creation process by AI methods that make suggestions and ask for clarifications, but that leaves final interpretation of those notes in the hands of human beings. The AI methods are the same as in fully automated systems, but how they are used is crucially different: rather than trying to replace humans, the goal is to assist them. This approach provides a cost-effective, accurate documentation service for physicians.
Read the full post here.