Table 7.

Performance of the rule-based NLP algorithm and machine learning models.

SensitivityDaytime sleepinessNappingNight wakingsSleep problemBad sleep qualitySnoringSleep duration
Specificity
F1
PPV
AUROC
Rule-based NLP1.000.501.000.850.620.941.00
1.000.990.990.930.510.971.00
1.000.980.990.910.910.971.00
1.000.500.750.800.600.891.00
1.000.970.980.920.500.86
DT0.860.890.820.780.630.750.79
0.860.890.800.720.570.750.74
0.900.980.810.740.580.750.76
0.860.890.840.840.810.780.77
0.850.890.790.710.560.75
LR0.900.470.920.910.420.420.77
0.770.500.830.580.500.500.67
0.810.480.860.590.450.460.71
0.890.940.910.820.830.840.78
0.780.500.820.580.500.50
KNN0.930.790.870.790.760.700.84
0.850.790.790.650.720.660.89
0.880.790.820.680.740.710.86
0.930.950.890.830.860.810.87
0.850.790.780.640.720.66
SVM0.910.810.900.910.930.950.85
0.800.690.850.610.600.690.84
0.840.740.870.630.630.750.84
0.900.950.910.830.860.900.78
0.790.690.850.610.600.68
LLAMA2-CoT0.690.720.480.820.740.800.77
0.540.620.340.860.670.850.72
0.570.580.390.830.700.790.67
0.540.620.340.860.670.840.72
0.540.600.370.840.680.78
LLAMA2-SFT0.930.820.900.900.870.881.00
0.920.940.940.900.870.881.00
0.910.880.960.890.840.781.00
0.920.850.930.890.840.831.00
0.910.820.900.890.870.87
SensitivityDaytime sleepinessNappingNight wakingsSleep problemBad sleep qualitySnoringSleep duration
Specificity
F1
PPV
AUROC
Rule-based NLP1.000.501.000.850.620.941.00
1.000.990.990.930.510.971.00
1.000.980.990.910.910.971.00
1.000.500.750.800.600.891.00
1.000.970.980.920.500.86
DT0.860.890.820.780.630.750.79
0.860.890.800.720.570.750.74
0.900.980.810.740.580.750.76
0.860.890.840.840.810.780.77
0.850.890.790.710.560.75
LR0.900.470.920.910.420.420.77
0.770.500.830.580.500.500.67
0.810.480.860.590.450.460.71
0.890.940.910.820.830.840.78
0.780.500.820.580.500.50
KNN0.930.790.870.790.760.700.84
0.850.790.790.650.720.660.89
0.880.790.820.680.740.710.86
0.930.950.890.830.860.810.87
0.850.790.780.640.720.66
SVM0.910.810.900.910.930.950.85
0.800.690.850.610.600.690.84
0.840.740.870.630.630.750.84
0.900.950.910.830.860.900.78
0.790.690.850.610.600.68
LLAMA2-CoT0.690.720.480.820.740.800.77
0.540.620.340.860.670.850.72
0.570.580.390.830.700.790.67
0.540.620.340.860.670.840.72
0.540.600.370.840.680.78
LLAMA2-SFT0.930.820.900.900.870.881.00
0.920.940.940.900.870.881.00
0.910.880.960.890.840.781.00
0.920.850.930.890.840.831.00
0.910.820.900.890.870.87

Highlighted are the best performances on each sleep concept.

Table 7.

Performance of the rule-based NLP algorithm and machine learning models.

SensitivityDaytime sleepinessNappingNight wakingsSleep problemBad sleep qualitySnoringSleep duration
Specificity
F1
PPV
AUROC
Rule-based NLP1.000.501.000.850.620.941.00
1.000.990.990.930.510.971.00
1.000.980.990.910.910.971.00
1.000.500.750.800.600.891.00
1.000.970.980.920.500.86
DT0.860.890.820.780.630.750.79
0.860.890.800.720.570.750.74
0.900.980.810.740.580.750.76
0.860.890.840.840.810.780.77
0.850.890.790.710.560.75
LR0.900.470.920.910.420.420.77
0.770.500.830.580.500.500.67
0.810.480.860.590.450.460.71
0.890.940.910.820.830.840.78
0.780.500.820.580.500.50
KNN0.930.790.870.790.760.700.84
0.850.790.790.650.720.660.89
0.880.790.820.680.740.710.86
0.930.950.890.830.860.810.87
0.850.790.780.640.720.66
SVM0.910.810.900.910.930.950.85
0.800.690.850.610.600.690.84
0.840.740.870.630.630.750.84
0.900.950.910.830.860.900.78
0.790.690.850.610.600.68
LLAMA2-CoT0.690.720.480.820.740.800.77
0.540.620.340.860.670.850.72
0.570.580.390.830.700.790.67
0.540.620.340.860.670.840.72
0.540.600.370.840.680.78
LLAMA2-SFT0.930.820.900.900.870.881.00
0.920.940.940.900.870.881.00
0.910.880.960.890.840.781.00
0.920.850.930.890.840.831.00
0.910.820.900.890.870.87
SensitivityDaytime sleepinessNappingNight wakingsSleep problemBad sleep qualitySnoringSleep duration
Specificity
F1
PPV
AUROC
Rule-based NLP1.000.501.000.850.620.941.00
1.000.990.990.930.510.971.00
1.000.980.990.910.910.971.00
1.000.500.750.800.600.891.00
1.000.970.980.920.500.86
DT0.860.890.820.780.630.750.79
0.860.890.800.720.570.750.74
0.900.980.810.740.580.750.76
0.860.890.840.840.810.780.77
0.850.890.790.710.560.75
LR0.900.470.920.910.420.420.77
0.770.500.830.580.500.500.67
0.810.480.860.590.450.460.71
0.890.940.910.820.830.840.78
0.780.500.820.580.500.50
KNN0.930.790.870.790.760.700.84
0.850.790.790.650.720.660.89
0.880.790.820.680.740.710.86
0.930.950.890.830.860.810.87
0.850.790.780.640.720.66
SVM0.910.810.900.910.930.950.85
0.800.690.850.610.600.690.84
0.840.740.870.630.630.750.84
0.900.950.910.830.860.900.78
0.790.690.850.610.600.68
LLAMA2-CoT0.690.720.480.820.740.800.77
0.540.620.340.860.670.850.72
0.570.580.390.830.700.790.67
0.540.620.340.860.670.840.72
0.540.600.370.840.680.78
LLAMA2-SFT0.930.820.900.900.870.881.00
0.920.940.940.900.870.881.00
0.910.880.960.890.840.781.00
0.920.850.930.890.840.831.00
0.910.820.900.890.870.87

Highlighted are the best performances on each sleep concept.

Close
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close

This PDF is available to Subscribers Only

View Article Abstract & Purchase Options

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Close