Table 3.

Model Utility in Health Care Delivery

Model performance	AUC, Sensitivity, Specificity, PPV, etc.
Data requirements	Shorter observation window, fewer features
Model coherence and translation	Transparency, interpretability, clinical and etiological utility, actionability
Model purpose and validation	Diagnosis, prognosis, risk management, risk communication, treatment/intervention indication
Ease of updating	Ease of implementation, platform utilities
Ease of use/access	User access, requirements, and burden; Integration with workflow