J Clin Med Res

Journal of Clinical Medicine Research, ISSN 1918-3003 print, 1918-3011 online, Open Access

Article copyright, the authors; Journal compilation copyright, J Clin Med Res and Elmer Press Inc

Journal website https://jocmr.elmerjournals.com

Original Article

Volume 18, Number 7, July 2026, pages 472-488

Development and Validation of an XGBoost-Based Machine Learning Model With Nomogram for Predicting Diabetic Peripheral Neuropathy Risk in Type 2 Diabetes Patients

↓ Figure 2. LASSO regression coefficient path diagram (a) and cross-validation error curve (b). The red dashed line indicates the optimal λ = 0.042, which corresponds to retaining eight non-zero coefficient features. LASSO: least absolute shrinkage and selection operator.

↓ Figure 4. SHAP analysis results. (a) SHAP heat map displays the global importance ranking of each feature and the direction of their impact (red = high feature value, blue = low feature value, positive x-axis values = increased DPN risk); (b) SHAP dependence plots for HbA1c and MAU show non-linear effects and threshold phenomena; (c) SHAP waterfall plot for individualized risk stratifications in a single patient example (high-risk patient case). DPN: diabetic peripheral neuropathy; HbA1c: glycated hemoglobin; MAU: microalbuminuria; SHAP: SHapley Additive exPlanations.

↓ Figure 5. A DPN risk stratification nomogram constructed based on six independent predictive factors using multivariate logistic regression. DPN: diabetic peripheral neuropathy.

↓ **Table 1.** Comparison of Baseline Characteristics Between DPN and Non-DPN Groups
Variables	Non-DPN group (n = 92)	DPN group (n = 88)	P-value
*P < 0.05. Independent sample t-test or Mann–Whitney U test was used for continuous variables and χ² test for categorical variables. ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BUN: blood urea nitrogen; DPN: diabetic peripheral neuropathy; eGFR: estimated glomerular filtration rate; FPG: fasting plasma glucose; HbA1c: glycated hemoglobin; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; IQR: interquartile range; LDL-C: low-density lipoprotein cholesterol; LR: logistic regression; MAU: microalbuminuria; NEUT%: neutrophil percentage; PLT: platelet count; SCr: serum creatinine; SD: standard deviation; TC: total cholesterol; TG: triglycerides.
Demographic characteristics
Age (years), mean ± SD	56.8 ± 11.2	61.4 ± 10.6	0.006*
Male, n (%)	52 (56.5)	54 (61.4)	0.505
Duration of diabetes (years), mean (IQR)	5.2 (2.5–9.0)	9.5 (5.0–15.0)	< 0.001*
Clinical measurements
BMI (kg/m²), mean ± SD	25.4 ± 3.2	26.8 ± 3.6	0.008*
Blood glucose metabolism
FPG (mmol/L), mean ± SD	8.6 ± 2.8	9.8 ± 3.1	0.008*
HbA1c (%), mean ± SD	7.8 ± 1.4	9.2 ± 1.8	< 0.001*
Lipid profile
TC (mmol/L), mean ± SD	4.52 ± 1.02	4.78 ± 1.18	0.124
TG (mmol/L), mean (IQR)	1.62 (1.12–2.38)	1.98 (1.28–2.85)	0.035*
HDL-C (mmol/L), mean ± SD	1.22 ± 0.30	1.08 ± 0.28	0.002*
LDL-C (mmol/L), mean ± SD	2.68 ± 0.82	3.05 ± 0.96	0.007*
Hepatic function
ALT (U/L), mean (IQR)	22.0 (15.0–35.0)	24.5 (16.0–40.0)	0.326
AST (U/L), mean (IQR)	20.0 (14.0–28.0)	22.0 (16.0–32.0)	0.213
Renal function
SCr (µmol/L), mean (IQR)	68.2 (56.4–82.6)	78.5 (62.0–98.4)	0.004*
BUN (mmol/L), mean ± SD	5.8 ± 1.6	6.5 ± 2.0	0.013*
eGFR (mL/min/1.73 m²), mean ± SD	88.4 ± 18.2	78.6 ± 20.5	0.001*
Urine parameters
MAU (mg/24 h), mean (IQR)	28.6 (14.2–58.5)	89.5 (38.6–185.2)	< 0.001*
Blood routine
HGB (g/L), mean ± SD	136.4 ± 17.2	128.5 ± 18.6	0.005*
PLT (× 10⁹/L), mean ± SD	218.5 ± 58.6	228.4 ± 65.2	0.299
NEUT% (%), mean ± SD	63.2 ± 8.5	68.4 ± 9.2	< 0.001*
Hypoglycemic medication, n (%)
Single oral hypoglycemic drugs	34 (37.0)	21 (23.9)	0.062
Combined oral hypoglycemic drugs	29 (31.5)	24 (27.3)	0.564
Insulin monotherapy	16 (17.4)	23 (26.1)	0.168
Insulin + oral hypoglycemic drugs	13 (14.1)	20 (22.7)	0.139
DPN severity, n (%)
Mild	-	42 (47.7)	-
Moderate	-	33 (37.5)	-
Severe	-	13 (14.8	-

↓ Table 1. Comparison of Baseline Characteristics Between DPN and Non-DPN Groups

Variables

Non-DPN group (n = 92)

DPN group (n = 88)

P-value

*P < 0.05. Independent sample t-test or Mann–Whitney U test was used for continuous variables and χ² test for categorical variables. ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BUN: blood urea nitrogen; DPN: diabetic peripheral neuropathy; eGFR: estimated glomerular filtration rate; FPG: fasting plasma glucose; HbA1c: glycated hemoglobin; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; IQR: interquartile range; LDL-C: low-density lipoprotein cholesterol; LR: logistic regression; MAU: microalbuminuria; NEUT%: neutrophil percentage; PLT: platelet count; SCr: serum creatinine; SD: standard deviation; TC: total cholesterol; TG: triglycerides.

Demographic characteristics

Age (years), mean ± SD

56.8 ± 11.2

61.4 ± 10.6

0.006*

Male, n (%)

52 (56.5)

54 (61.4)

0.505

Duration of diabetes (years), mean (IQR)

5.2 (2.5–9.0)

9.5 (5.0–15.0)

< 0.001*

Clinical measurements

BMI (kg/m²), mean ± SD

25.4 ± 3.2

26.8 ± 3.6

0.008*

Blood glucose metabolism

FPG (mmol/L), mean ± SD

8.6 ± 2.8

9.8 ± 3.1

0.008*

HbA1c (%), mean ± SD

7.8 ± 1.4

9.2 ± 1.8

< 0.001*

Lipid profile

TC (mmol/L), mean ± SD

4.52 ± 1.02

4.78 ± 1.18

0.124

TG (mmol/L), mean (IQR)

1.62 (1.12–2.38)

1.98 (1.28–2.85)

0.035*

HDL-C (mmol/L), mean ± SD

1.22 ± 0.30

1.08 ± 0.28

0.002*

LDL-C (mmol/L), mean ± SD

2.68 ± 0.82

3.05 ± 0.96

0.007*

Hepatic function

ALT (U/L), mean (IQR)

22.0 (15.0–35.0)

24.5 (16.0–40.0)

0.326

AST (U/L), mean (IQR)

20.0 (14.0–28.0)

22.0 (16.0–32.0)

0.213

Renal function

SCr (µmol/L), mean (IQR)

68.2 (56.4–82.6)

78.5 (62.0–98.4)

0.004*

BUN (mmol/L), mean ± SD

5.8 ± 1.6

6.5 ± 2.0

0.013*

eGFR (mL/min/1.73 m²), mean ± SD

88.4 ± 18.2

78.6 ± 20.5

0.001*

Urine parameters

MAU (mg/24 h), mean (IQR)

28.6 (14.2–58.5)

89.5 (38.6–185.2)

< 0.001*

Blood routine

HGB (g/L), mean ± SD

136.4 ± 17.2

128.5 ± 18.6

0.005*

PLT (× 10⁹/L), mean ± SD

218.5 ± 58.6

228.4 ± 65.2

0.299

NEUT% (%), mean ± SD

63.2 ± 8.5

68.4 ± 9.2

< 0.001*

Hypoglycemic medication, n (%)

Single oral hypoglycemic drugs

34 (37.0)

21 (23.9)

0.062

Combined oral hypoglycemic drugs

29 (31.5)

24 (27.3)

0.564

Insulin monotherapy

16 (17.4)

23 (26.1)

0.168

Insulin + oral hypoglycemic drugs

13 (14.1)

20 (22.7)

0.139

DPN severity, n (%)

Mild

42 (47.7)

Moderate

33 (37.5)

Severe

13 (14.8

↓ **Table 2.** Univariate Logistic Regression Analysis of Risk Factors for DPN
Variables	β	OR	95% CI	P-value
^aMAU was included in regression analysis (ln-MAU) after natural logarithmic transformation. *P < 0.05 was included in multivariate analysis candidate set. ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BUN: blood urea nitrogen; CI: confidence interval; DPN: diabetic peripheral neuropathy; eGFR: estimated glomerular filtration rate; FPG: fasting plasma glucose; HbA1c: glycated hemoglobin; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; IQR: interquartile range; LDL-C: low-density lipoprotein cholesterol; LR: logistic regression; MAU: microalbuminuria; NEUT%: neutrophil percentage; OR: odds ratio; PLT: platelet count; SCr: serum creatinine; SD: standard deviation; TC: total cholesterol; TG: triglycerides.
Age (years)	0.042	1.043	1.012–1.075	0.007*
Gender (male vs. female)	0.182	1.200	0.614–2.344	0.592
Duration of diabetes (years)	0.148	1.160	1.095–1.228	< 0.001*
BMI (kg/m²)	0.128	1.137	1.036–1.247	0.007*
FPG (mmol/L)	0.138	1.148	1.042–1.264	0.005*
HbA1c (%)	0.524	1.689	1.386–2.057	< 0.001*
TC (mmol/L)	0.218	1.243	0.938–1.647	0.130
TG (mmol/L)	0.288	1.334	1.038–1.714	0.025*
HDL-C (mmol/L)	−1.648	0.192	0.082–0.451	< 0.001*
LDL-C (mmol/L)	0.442	1.556	1.164–2.080	0.003*
ALT (U/L)	0.008	1.008	0.993–1.024	0.282
AST (U/L)	0.012	1.012	0.993–1.032	0.226
SCr (µmol/L)	0.012	1.012	1.002–1.022	0.014*
BUN (mmol/L)	0.184	1.202	1.036–1.395	0.015*
eGFR (mL/min/1.73 m²)	−0.022	0.978	0.963–0.993	0.004*
MAU (mg/24 h)^a	0.682	1.978	1.512–2.587	< 0.001*
HGB (g/L)	−0.018	0.982	0.968–0.997	0.018*
PLT (× 10⁹/L)	0.003	1.003	0.997–1.009	0.322
NEUT% (%)	0.068	1.070	1.033–1.109	< 0.001*

↓ Table 2. Univariate Logistic Regression Analysis of Risk Factors for DPN

Variables

95% CI

P-value

^aMAU was included in regression analysis (ln-MAU) after natural logarithmic transformation. *P < 0.05 was included in multivariate analysis candidate set. ALT: alanine aminotransferase; AST: aspartate aminotransferase; BMI: body mass index; BUN: blood urea nitrogen; CI: confidence interval; DPN: diabetic peripheral neuropathy; eGFR: estimated glomerular filtration rate; FPG: fasting plasma glucose; HbA1c: glycated hemoglobin; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; IQR: interquartile range; LDL-C: low-density lipoprotein cholesterol; LR: logistic regression; MAU: microalbuminuria; NEUT%: neutrophil percentage; OR: odds ratio; PLT: platelet count; SCr: serum creatinine; SD: standard deviation; TC: total cholesterol; TG: triglycerides.

Age (years)

0.042

1.043

1.012–1.075

0.007*

Gender (male vs. female)

0.182

1.200

0.614–2.344

0.592

Duration of diabetes (years)

0.148

1.160

1.095–1.228

< 0.001*

BMI (kg/m²)

0.128

1.137

1.036–1.247

0.007*

FPG (mmol/L)

0.138

1.148

1.042–1.264

0.005*

HbA1c (%)

0.524

1.689

1.386–2.057

< 0.001*

TC (mmol/L)

0.218

1.243

0.938–1.647

0.130

TG (mmol/L)

0.288

1.334

1.038–1.714

0.025*

HDL-C (mmol/L)

−1.648

0.192

0.082–0.451

< 0.001*

LDL-C (mmol/L)

0.442

1.556

1.164–2.080

0.003*

ALT (U/L)

0.008

1.008

0.993–1.024

0.282

AST (U/L)

0.012

1.012

0.993–1.032

0.226

SCr (µmol/L)

0.012

1.012

1.002–1.022

0.014*

BUN (mmol/L)

0.184

1.202

1.036–1.395

0.015*

eGFR (mL/min/1.73 m²)

−0.022

0.978

0.963–0.993

0.004*

MAU (mg/24 h)^a

0.682

1.978

1.512–2.587

< 0.001*

HGB (g/L)

−0.018

0.982

0.968–0.997

0.018*

PLT (× 10⁹/L)

0.003

1.003

0.997–1.009

0.322

NEUT% (%)

0.068

1.070

1.033–1.109

< 0.001*

↓ **Table 3.** Multivariate Logistic Regression Analysis of Independent Risk Factors for DPN
Independent predictors	β	OR	95% CI	P-value	VIF
BMI: body mass index; CI: confidence interval; DPN: diabetic peripheral neuropathy; HbA1c: glycated hemoglobin; LDL-C: low-density lipoprotein cholesterol; MAU: microalbuminuria; NEUT%: neutrophil percentage; OR: odds ratio.
Duration of diabetes (years)	0.118	1.125	1.052–1.203	0.001	1.42
HbA1c (%)	0.406	1.501	1.196–1.883	< 0.001	1.68
ln-MAU (mg/24 h)	0.558	1.747	1.286–2.373	< 0.001	1.35
LDL-C (mmol/L)	0.382	1.465	1.062–2.021	0.020	1.28
NEUT% (%)	0.056	1.058	1.016–1.101	0.006	1.22
BMI (kg/m²)	0.102	1.107	1.012–1.212	0.026	1.18
Constant term	−16.428	-	-	< 0.001	-

↓ Table 3. Multivariate Logistic Regression Analysis of Independent Risk Factors for DPN

Independent predictors

95% CI

P-value

VIF

BMI: body mass index; CI: confidence interval; DPN: diabetic peripheral neuropathy; HbA1c: glycated hemoglobin; LDL-C: low-density lipoprotein cholesterol; MAU: microalbuminuria; NEUT%: neutrophil percentage; OR: odds ratio.

Duration of diabetes (years)

0.118

1.125

1.052–1.203

0.001

1.42

HbA1c (%)

0.406

1.501

1.196–1.883

< 0.001

1.68

ln-MAU (mg/24 h)

0.558

1.747

1.286–2.373

< 0.001

1.35

LDL-C (mmol/L)

0.382

1.465

1.062–2.021

0.020

1.28

NEUT% (%)

0.056

1.058

1.016–1.101

0.006

1.22

BMI (kg/m²)

0.102

1.107

1.012–1.212

0.026

1.18

Constant term

−16.428

< 0.001

↓ **Table 4.** Performance of 10-Fold Cross-Validation for Five Machine Learning Models
Models	AUC (mean ± SD)	Accuracy (%)	Sensitivity (%)	Specificity (%)	F1 score
AUC: area under the curve; SD: standard deviation; SVM: support vector machine.
XGBoost	0.912 ± 0.048	85.7	84.2	87.2	0.858
Random forest	0.896 ± 0.052	83.3	82.6	84.0	0.836
Logistic regression	0.868 ± 0.058	80.2	79.5	81.0	0.802
SVM	0.845 ± 0.062	78.6	77.8	79.5	0.784
Decision tree	0.812 ± 0.072	76.2	75.4	77.0	0.760

↓ Table 4. Performance of 10-Fold Cross-Validation for Five Machine Learning Models

Models

AUC (mean ± SD)

Accuracy (%)

Sensitivity (%)

Specificity (%)

F1 score

AUC: area under the curve; SD: standard deviation; SVM: support vector machine.

XGBoost

0.912 ± 0.048

85.7

84.2

87.2

0.858

Random forest

0.896 ± 0.052

83.3

82.6

84.0

0.836

Logistic regression

0.868 ± 0.058

80.2

79.5

81.0

0.802

SVM

0.845 ± 0.062

78.6

77.8

79.5

0.784

Decision tree

0.812 ± 0.072

76.2

75.4

77.0

0.760

↓ **Table 5.** Performance Comparison of Five Machine Learning Models on the Validation Set
Models	AUC (95% CI)	Accuracy (%)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	F1 score	TP	FP	TN	FN
AUC: area under the curve; CI: confidence interval; FN: false negatives; FP: false positives; NPV: negative predictive value; PPV: positive predictive value; SVM: support vector machine; TN: true negatives; TP: true positives.
XGBoost	0.903 (0.816–0.986)	85.2	84.6	85.7	84.6	85.7	0.846	22	4	24	4
Random forest	0.808 (0.713–0.895)	83.3	80.8	85.7	84.0	82.8	0.824	20	5	23	6
Logistic regression	0.838 (0.755–0.957)	79.6	80.8	78.6	77.8	81.5	0.793	19	6	22	7
SVM	0.884 (0.726–0.988)	77.8	76.9	78.6	76.9	78.6	0.769	21	5	23	5
Decision tree	0.814 (0.683–0.933)	74.1	73.1	75.0	73.1	75.0	0.731	18	7	21	8

↓ Table 5. Performance Comparison of Five Machine Learning Models on the Validation Set

Models

AUC (95% CI)

Accuracy (%)

Sensitivity (%)

Specificity (%)

PPV (%)

NPV (%)

F1 score

AUC: area under the curve; CI: confidence interval; FN: false negatives; FP: false positives; NPV: negative predictive value; PPV: positive predictive value; SVM: support vector machine; TN: true negatives; TP: true positives.

XGBoost

0.903 (0.816–0.986)

85.2

84.6

85.7

84.6

85.7

0.846

Random forest

0.808 (0.713–0.895)

83.3

80.8

85.7

84.0

82.8

0.824

Logistic regression

0.838 (0.755–0.957)

79.6

80.8

78.6

77.8

81.5

0.793

SVM

0.884 (0.726–0.988)

77.8

76.9

78.6

76.9

78.6

0.769

Decision tree

0.814 (0.683–0.933)

74.1

73.1

75.0

73.1

75.0

0.731