Skip to main content

Table 6 Comparison of AI models with ASA-PS for primary composite adverse outcomes, ICU admission and prolonged length of hospital stay

From: Implementation of a machine learning application in preoperative risk assessment for hip repair surgery

Outcome

Model

Accuracy

Sensitivity

Specificity

AUC (95%CI)

P-value%

Composite adverse

ASA

0.326

0.896

0.262

0.629 (0.590–0.668)

 < 0.001

bAI model

0.538

0.903

0.529

0.794 (0.718–0.869)

ICU admission

ASA

0.931

0.240

0.958

0.692 (0.645–0.738)

 < 0.001

bAI model

0.979

0.240

0.979

0.856 (0.804–0.908)

aPLOS

ASA

0.336

0.909

0.279

0.618 (0.582–0.654)

 < 0.001

cAI model

0.649

0.908

0.624

0.854 (0.818–0.890)

  1. a PLOS Prolong length of hospital-stay
  2. b Using logistic regression for AI models for primary composite adverse outcomes and ICU admission
  3. c Using Random Forest for AI model for PLOS
  4. % Delong test