Table 3 Performance of the gauze EBL regression models on the human patient dataset.

From: Automated deep learning model for estimating intraoperative blood loss using gauze images

 

P-W CS model

MAE (g)

MAPE (%)

Performance of the P-W CS model evaluated on the human patient dataset (n = 473)

 EBL value (x)

\(\mathrm{ x}\le 3{\text{g}}\)

0.21

15.73

  \(3{\text{g}}<{\text{x}}\le 5{\text{g}}\)

0.22

5.27

  \({\text{x}}>5{\text{g}}\)

0.42

7.67

 Average

0.25

7.26

  1. EBL, Estimated blood loss; CS, Crumpled state; P-W CS, Patch-wise crumpled state; MAE, Mean absolute error; MAPE, Mean absolute percentage error.
  2. Significant values are in bold.