What is the pooled AUC for CAD prediction?

The prediction in CAD was associated with pooled AUC of 0.87 (95% CI 0.76–0.93) for CNN, pooled AUC of 0.88 (95% CI 0.84–0.91) for boosting algorithms, and pooled of AUC 0.93 (95% CI 0.85–0.97) for others (custom-built algorithms).

Should ML algorithms be used to predict CAD instances?

Ideally, for healthy subjects, this procedure should be avoided. There is a plethora of related work 4, 5, 6, 7, 8, 9, 10 where common ML algorithms are used to predict CAD instances with varying results, ranging accuracy-wise from 71.1% to over 98% when also employing image data.

Can human expertise improve the diagnosis of CAD?

The results of this study demonstrate the potential for this approach to improve the diagnosis of CAD and highlight the importance of considering the role of human expertise in the development of computer-aided classification models.

What are the CAD predicting variables?

Based on the calculated weights, the following variables were selected as CAD predicting variables: gender, occupation, place of residence, family history, smoking status, comorbidity, mean value of pulse rate, TST waves status, hypertension history, chest pain, cholesterol, triglyceride, blood glucose level and creatinine level.