Skip to main content

Table 4 Performance of machine learning approach after error analysis

From: Machine learning algorithms for systematic review: reducing workload in a preclinical review of animal studies and reducing human screening error

  Updated approach 1 Original approach 1
Cut-off 0.09 0.10
Sensitivity 98.7% 98.7%
Upper 95% CI of sensitivity 0.997 0.997
Lower 95% CI of sensitivity 0.949 0.949
Specificity 88.3% 86.7%
Precision 55.9% 52.61%
Accuracy 89.7% 88.2%
WSS@95% 961/1251 – (0.05) = 0.718 945/1251 – (0.05) = 0.705
LR+ 8.436 7.421