Table 4 Best performing models for the various biomarkers in both male and female subgroups as well as combined data and combined data with sex as an input feature

From: A supervised machine learning approach with feature selection for sex-specific biomarker prediction

 

Prediction Target

Model

MAE (Unit)

MSE (Unit²)

RMSE (Unit)

RMSLE

MAPE

Female

Waist circumference (cm)

Bayesian Ridge

4.95 cm

40.55 cm²

6.37 cm

0.8

0.07

0.05

BMI (kg/m²)

Gradient Boosting Regressor

2.05 kg/m²

7.54 kg/m²

2.75 kg/m²

0.77

0.09

0.07

UrAlbCr (mg/g)

Linear Regression

3.11 mg/g

16.51 mg/g²

4.06 mg/g

-0.01

0.46

0.53

Uric Acid (mmol/L)

Huber Regressor

0.03 mmol/L

0.00 mmol/L²

0.04 mmol/L

0.11

0.04

0.19

Blood Glucose (mmol/L)

Linear Regression

0.47 mmol/L

0.36 mmol/L²

0.60 mmol/L

0.17

0.09

0.09

HDL (mmol/L)

Bayesian Ridge

0.21 mmol/L

0.07 mmol/L²

0.26 mmol/L

0.23

0.1

0.15

Triglycerides (mmol/L)

Linear Regression

0.33 mmol/L

0.21 mmol/L²

0.46 mmol/L

0.21

0.2

0.33

Albuminuria (mg/L)

Bayesian Ridge

0.19 mg/L

0.06 mg/L²

0.24 mg/L

0.2

0.05

0.05

Systolic Blood Pressure (mmHg)

Huber Regressor

10.37 mmHg

175.94 mmHg²

13.26 mmHg

0.24

0.11

0.09

Male

Waist circumference (cm)

Huber Regressor

3.90 cm

25.39 cm²

5.04 cm

0.86

0.05

0.04

BMI (kg/m²)

Ridge

1.43 kg/m²

3.15 kg/m²

1.77 kg/m²

0.86

0.06

0.05

UrAlbCr (mg/g)

Random Forest Regressor

2.84 mg/g

19.30 mg/g²

4.39 mg/g

0

0.49

0.46

Uric Acid (mmol/L)

Bayesian Ridge

0.03 mmol/L

0.00 mmol/L²

0.04 mmol/L

0.12

0.04

0.17

Blood Glucose (mmol/L)

Huber Regressor

0.40 mmol/L

0.28 mmol/L²

0.53 mmol/L

0.14

0.08

0.07

HDL (mmol/L)

Linear Regression

0.20 mmol/L

0.07 mmol/L²

0.27 mmol/L

0.14

0.11

0.15

Triglycerides (mmol/L)

ExtraTrees Regressor

0.41 mmol/L

0.26 mmol/L²

0.51 mmol/L

0.13

0.21

0.34

Albuminuria (mg/L)

Bayesian Ridge

0.21 mg/L

0.07 mg/L²

0.26 mg/L

0.14

0.05

0.05

Systolic Blood Pressure (mmHg)

Huber Regressor

10.38 mmHg

176.90 mmHg²

13.30 mmHg

0.08

0.11

0.09

Combined Data excluding sex as an input feature

Waist circumference (cm)

Bayesian Ridge

4.16 cm

32.15 cm²

5.67 cm

0.83

0.06

0.04

BMI (kg/m²)

Gradient Boosting Regressor

1.76 kg/m²

5.47 kg/m²

2.34 kg/m²

0.81

0.08

0.06

UrAlbCr (mg/g)

Random Forest Regressor

2.85 mg/g

15.92 mg/g²

3.99 mg/g

0.058

0.46

0.50

Uric Acid (mmol/L)

BayesianRidge

0.03 mmol/L

0.00 mmol/L²

0.04 mmol/L

0.23

0.03

0.16

Blood Glucose (mmol/L)

Huber Regressor

0.43 mmol/L

0.30 mmol/L²

0.55 mmol/L

0.21

0.08

0.08

HDL (mmol/L)

LinearRegression

0.21 mmol/L

0.07 mmol/L²

0.27 mmol/L

0.27

0.11

0.16

Triglycerides (mmol/L)

LinearRegression

0.35 mmol/L

0.22 mmol/L²

0.47 mmol/L

0.20

0.20

0.35

Albuminuria (mg/L)

Ridge

0.21 mg/L

0.07 mg/L²

0.27 mg/L

0.15

0.05

0.05

Systolic Blood Pressure (mmHg)

Random Forest Regressor

10.77 mmHg

198.82 mmHg²

14.10 mmHg

0.13

0.11

0.09

Combined Data including sex as an input feature

Waist circumference (cm)

Bayesian Ridge

3.98 cm

28.10 cm²

5.30 cm

0.85

0.05

0.04

BMI (kg/m²)

Gradient Boosting Regressor

1.62 kg/m²

4.53 kg/m²

2.13 kg/m²

0.85

0.07

0.06

UrAlbCr (mg/g)

AdaBoost Regressor

2.78 mg/g

15.32 mg/g²

3.91 mg/g

0.09

0.45

0.51

Uric Acid (mmol/L)

Bayesian Ridge

0.03 mmol/L

0.00 mmol/L²

0.04 mmol/L

0.35

0.03

0.15

Blood Glucose (mmol/L)

Bayesian Ridge

0.43 mmol/L

0.30 mmol/L²

0.55 mmol/L

0.21

0.08

0.08

HDL (mmol/L)

Linear Regression

0.20 mmol/L

0.07 mmol/L²

0.26 mmol/L

0.34

0.11

0.15

Triglycerides (mmol/L)

Linear Regression

0.35 mmol/L

0.21 mmol/L²

0.46 mmol/L

0.23

0.20

0.34

Albuminuria (mg/L)

Bayesian Ridge

0.21 mg/L

0.07 mg/L²

0.27 mg/L

0.19

0.05

0.05

Systolic Blood Pressure (mmHg)

Random Forest Regressor

10.56 mmHg

194.87 mmHg²

13.96 mmHg

0.15

0.11

0.09