In the future, the evolving role of artificial intelligence will likely alter the language and methods used for improvement. Predictive analytics and machine learning are likely to create opportunities to detect errors as they evolve in real time, providing time to reconsider diagnostic conclusions. Novel data sources and analytic methods will create new opportunities to design better care, and as they do, they will likely add complexity and give rise to new terms to add to the ones discussed in this brief.48
In particular, distinctions about prospective versus retrospective judgments may be replaced by just-in-time predictive algorithms that are more proximate to the moment of care. It is worth examining these new methods to make sure the terminology we adopt continues to advance understanding and support progress in improvement, including recognition of potential harm. Just as seen with the implementation of electronic health records and computer physician order entry for medications, new technology provides novel solutions but also introduces new types of risk. These risks need to be anticipated, monitored, described, and managed. The taxonomy for diagnostic improvement will need to parallel these advancements.
