A study involving 15 participants shows that interactive risk prediction tools are better understood than expected, but that technical jargon and the way risk is framed remain obstacles.
Machine learning models that estimate a risk of disease are coming to tools aimed directly at patients. It remains to be seen whether patients understand what they are being shown. In an exploratory work published on arXiv, researchers studied how 15 participants interpret an interactive interface for predicting the risk of type 2 diabetes, and whether its use modifies their intention to change their habits.
By comparing perceived understanding and actual understanding, the team found that participants often understood the interface better than they thought. However, several obstacles remain: unclear terminology, an ambiguous presentation of risk and too limited explanations of the model's input data.
The authors draw design recommendations—intuitive visuals, familiar presentation, explicit explanations—and highlight the issues of trust and fairness that these tools pose.
Read the full article on the original source
Go to the source →