Predicting Diabetes in Patients with Metabolic Syndrome Using Machine-Learning Model Based on Multiple Years’ Data

Figures & data

Table 1 The Criteria of the International Diabetes Federation (IDF) for the Definition of Metabolic Syndrome (MetS)

The Patient With Mets Should Meet At Least Any Three Of The Following Factors
Waist circumference	Male≥90 cm; Female≥80 cm
Triglyceride	≥1.7 mmol/L
High-density lipoprotein cholesterol	Male<1.03 mmol/L; Female<1.29 mmol/L
Blood pressure	Systolic blood pressure≥130 mmHg or Diastolic blood pressure≥85 mmHg or Medical history with high blood pressure
Fasting plasma glucose	≥5.6 mmol/L or medical history with diabetes

Table 2 The Missing Percentages of Each Variable

Variables	Missing Percentage (%)
Age	0
Height (cm)	0.43
Weight (kg)	0.43
BMI (kg/m^2)	0.43
WC (cm)	41.10
SBP (mmHg)	0.46
DBP (mmHg)	0.47
HDLC (mmol/L)	6.25
LDLC (mmol/L)	2.33
TG (mmol/L)	0.53
FPG (mmol/L)	0.62
HbA1c (mmol/L)	38.68
TSH (mmol/L)	35.56
UA (mmol/L)	3.75

Abbreviations: BMI, body mass index; WC, waist circumference; SBP, systolic blood pressure. DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDLC, high-density lipoprotein cholesterol; LDLC, low-density lipoprotein cholesterol; TG, triglyceride; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; TSH, thyroid-stimulating hormone; UA, uric acid.

Figure 1 The definition of each longitudinal dataset in the timeline.

Table 3 Baseline Characteristics of Sub-Groups from Patient Cohorts

Variables	Patients With Diabetes	Patients Without Diabetes	P-value
	Mean±SD	Mean±SD
Age	55.71±14.60	47.06±13.01	^b
Height (cm)	167.22±9.15	168.94±8.91	^b
Weight (kg)	76.54±13.07	73.83±11.80	^b
BMI (kg/m^2)	27.25±3.33	25.75±2.96	^b
WC (cm)	91.46±9.31	87.43±8.52	^b
SBP (mmHg)	135.27±17.80	125.52±15.44	^b
DBP (mmHg)	79.55±10.73	76.76±9.82	^b
HDLC (mmol/L)	1.13±0.27	1.16±0.26	^b
LDLC (mmol/L)	3.20±0.85	3.19±0.77	^a
TG (mmol/L)	2.22±2.22	1.90±1.36	^b
FPG (mmol/L)	6.43±1.11	5.37±0.45	^b
HbA1c (mmol/L)	6.00±0.65	5.47±0.30	^b
TSH (mmol/L)	2.61±3.61	2.34±2.43	^b
UA (mmol/L)	364.85±78.98	354.98±84.40	^b

Notes: ^aRepresents P-value<0.05, ^bRepresents P-value<0.01.

Abbreviations: BMI, body mass index; WC, waist circumference; SBP, systolic blood pressure. DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDLC, high-density lipoprotein cholesterol; LDLC, low-density lipoprotein cholesterol; TG, triglyceride; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; TSH, thyroid-stimulating hormone; UA, uric acid.

Table 4 Performance Metrics of Machine-Learning Models Using Longitudinal Data

Models	AUROC	Recall	Precision	F1-Score
Year 1
Logistic Regression	0.794	0.681	0.728	0.702
Random Forest	0.804	0.747	0.737	0.738
Xgboost	0.772	0.687	0.719	0.699
Year 2
Logistic Regression	0.838	0.746	0.783	0.763
Random Forest	0.838	0.747	0.752	0.748
Xgboost	0.823	0.759	0.757	0.756
Year 3
Logistic Regression	0.828	0.728	0.775	0.748
Random Forest	0.862	0.774	0.778	0.766
Xgboost	0.833	0.789	0.798	0.789
Year 1–2
Logistic Regression	0.670	0.549	0.646	0.584
Random Forest	0.847	0.801	0.794	0.796
Xgboost	0.856	0.748	0.779	0.757
Year 1–3
Logistic Regression	0.686	0.597	0.622	0.603
Random Forest	0.893	0.789	0.820	0.803
Xgboost	0.897	0.831	0.837	0.834

Abbreviation: AUROC, area under the receiver operating characteristic curve.

Figure 2 ROC curves of the three models for all the datasets.

Notes: (A) the ROC curve of logistic regression for single-year models and multiple-year models; (B) the ROC curve of random forest for single-year models and multiple-year models; (C) the ROC curve of Xgboost for single-year models and multiple-year models.

Abbreviation: ROC, receiver operating characteristic.

Figure 3 Feature importance of each dataset using LASSO.

Notes: Parameter “Del_xx” was abbreviated from “delta_xx”. “Var_xx” was abbreviated from “Variance_xx”.

Abbreviations: BMI, body mass index; WC, waist circumference; SBP, systolic blood pressure. DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDLC, high-density lipoprotein cholesterol; LDLC, low-density lipoprotein cholesterol; TG, triglyceride; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; TSH, thyroid-stimulating hormone; UA, uric acid.