Skip to main content

Table 5 Performance of the models on unforeseen test data

From: Leveraging machine learning for duration of surgery prediction in knee and hip arthroplasty – a development and validation study

Indication

Model

Mean absolute percentage error (%)

Root mean squared error (min)

Mean absolute error (min)

Hip (n = 271)

XGBoost

23.30 (20.89–25.92)

15.75 (13.53–18.02)

10.87 (9.76–12.05)

Linear multivariable regression

24.59 (21.83–27.61)

16.00 (13.65–18.49)

11.13 (10.02–12.29)

Knee (n = 184)

XGBoost

21.04 (18.49–23.87)

17.91 (13.96–22.01)

12.53 (11.04–14.15)

Linear multivariable regression

27.08 (23.33–31.22)

20.23 (16.06–24.4)

13.64 (11.89–15.52)

  1. 95% confidence intervals in parenthesis. Confidence intervals were derived using bootstrapping with 10,000 repetitions
  2. a) Hip arthroplasty