Greater attention to human factors and new techniques may change the way artificial intelligence is trained with small data, according to an article published by Harvard Business Review.
Researchers from Accenture wanted to see if opportunities lay within smaller data sets that go unused by organizations. For their study, the researchers focused on annotations added to medical charts by medical coders. With their tens of annotations on each of several thousand charts, the annotations are much smaller compared to data sets with a billion columns and rows.
In the experiment, the coders studied RNs who regularly used AI in their coding processes to link medical conditions with proper codes. The researchers wanted to find out how to transform the coders into AI trainers.
The 12-week study showed that close attention to human factors is needed to create and transform work processes through small data sets.
Three principles on human interaction with AI arose:
- "Balance machine learning with human domain expertise"
- "Focus on the quality of human input, not the quantity of machine output"
- "Recognize the social dynamics in play on teams working with small data"
To read the full article, click here.
More articles on healthcare finance:
Bankrupt California health system lays off 920 employees
Creditor seeks to oust hospital chain CEO
Trump budget proposal to cut Medicaid, CDC funding: 5 notes