Skip to main content

Table 2 Perfomance Measures

From: Predicting patient outcomes in psychiatric hospitals with routine data: a machine learning approach

 Sensitivity / RecallSpecificityPositive Predictive Value / PrecisionNegative Predictive ValuePrevalenceDetection PrevalenceBalanced Accuracy
Precision at least 20%
 Non-Response1.000.000.231.000.231.000.50
 Coercive Treatment0.730.780.200.970.070.260.76
 Long LOS0.980.280.200.990.160.760.63
 Short LOS0.830.370.200.920.160.660.60
Precision at least 25%
 Non-Response0.960.150.250.930.230.870.56
 Coercive Treatment0.480.890.250.960.070.130.69
 Long LOS0.940.480.250.980.160.580.71
 Short LOS0.610.650.250.900.160.390.63
Precision at least 33%
 Non-Response0.520.690.330.830.230.360.61
 Coercive Treatment0.230.970.330.940.070.050.60
 Long LOS0.490.820.330.900.160.230.65
 Short LOS0.410.840.330.880.160.200.62
  1. Outcomes without clinically meaningful operational points are not shown (Crisis Intervention & 1:1 Observation). Actual precision could be more than minimum precision. TP True Positive, FP False Positive, TN True negative, FN False Negative, Sensitivity = TP/(TP+ FN), Specificity = TN/(TN + FP), Positive Predictive Value = TP/(TP + FP), Negative Predictive Value = TN/(TN + FN), Prevalence = (TP + FN)/(TP + FP + TN + FN), Detection Prevalence = (TP + FP)/(TP + FP + TN + FN), Balanced Accuracy = (Sensitivity+Specificity)/2