Table 3 Logistic regression with feature selection performed by ChatGPT-4 for diabetic retinopathy using the training dataset

Age (years, per 1 unit increase)

−0.020

0.98 (0.97–0.99)

0.002

–

–

–

BMI (kg/m², per 1 unit increase)

−0.043

0.96 (0.93–0.99)

0.011

–

–

–

DM duration (years, per 1 unit increase)

0.038

1.04 (1.02–1.06)

 < 0.001

0.058

1.06 (1.04–1.10)

 < 0.001

DM oral medication

0.682

1.97 (1.37–2.44)

 < 0.001

–

–

–

DM insulin treatment

1.068

2.91 (1.93–4.75)

 < 0.001

–

–

–

SBP (mmHg, per 1 unit increase)

0.015

1.02 (1.00–1.03)

0.001

0.024

1.03 (1.01–1.05)

0.014

HbA1c (%, per 1 unit increase)

0.205

1.23 (1.11–1.39)

 < 0.001

0.534

1.70 (1.27–2.00)

 < 0.001

Fasting glucose (mg/dL, per 1 unit increase)

0.007

1.01 (1.00–1.01)

0.002

–

–

–

Creatinine (mg/dL, per 1 unit increase)

–

–

–

0.765

2.15 (1.38–3.03)

0.021

WBC (cells/μL, per 1 unit increase)

–

–

–

0.200

1.22 (1.02–1.51)

0.033

Platelet (10⁹/L, per 1 unit increase)

–

–

–

−0.006

0.99 (0.98–0.99)

0.025

Hemoglobin (mg/dL, per 1 unit increase)

−0.074

0.93 (0.86–0.99)

0.034

−0.366

0.79 (0.57–0.92)

0.007

Constant

−3.167

–

0.002

−7.731

–

 < 0.001