In the UK Biobank, combining three risk tools scored 0.811 for distinguishing future heart failure, versus 0.788 for clinical risk factors alone.In the UK Biobank, combining three risk tools scored 0.811 for distinguishing future heart failure, versus 0.788 for clinical risk factors alone.
Artificial intelligence beat a clinical risk-factor model for predicting one subtype, scoring 0.86 versus 0.78, but scored modestly lower for the other.Artificial intelligence beat a clinical risk-factor model for predicting one subtype, scoring 0.86 versus 0.78, but scored modestly lower for the other.
Who was studiedPeople without prior heart failure, 10 years, against a clinical risk-factor model
What changedPrediction scores were higher for reduced-ejection-fraction heart failure (0.86 versus 0.78) but lower for preserved-ejection-fraction heart failure (0.78 versus 0.81)
The catchThe evaluation measured prediction, not whether using predictions prevented heart failure
In Massachusetts General Hospital’s test group, 10-year prediction scores were 0.87 for heart failure with reduced ejection fraction and 0.79 for preserved ejection…In Massachusetts General Hospital’s test group, 10-year prediction scores were 0.87 for heart failure with reduced ejection fraction and 0.79 for preserved ejection fraction.