Table 1 Prompts used to develop a risk calculator to predict diabetic retinopathy and macular edema

1. Load the training dataset

(After dragging the training dataset into the dialog window) This data file is designed to predict diabetic retinopathy and diabetic macular edema using patients’ medical history and laboratory data

2. Develop DR prediction formula

Develop a formula to predict the occurrence of diabetic retinopathy using the following items

The variable targeted for prediction is E_DR (whether diabetic retinopathy occurs)

Please use logistic regression analysis using backward elimination feature selection method

Enter the following variables as input:

Sex, age, ho_incm, Dyslipidemia, DM_insulin, DM_duration, DM_po_med, Smoking, HE_sbp, HE_dbp, HE_wc, HE_BMI, HE_glu, HE_HbA1c, HE_chol, HE_TG, HE_ast, HE_alt, HE_HB, HE_BUN, HE_crea, HE_WBC, HE_Bplt, HE_Uacid

Please show the final prediction formula to predict E_DR

Draw the ROC curve of the E_DR prediction formula using the formula developed above. Show the optimal cutoff value

3. Develop DME prediction formula

Develop a formula to predict the occurrence of diabetic macular edema using the following items

The variable targeted for prediction is E_DME (whether diabetic macular edema occurs)

Please use logistic regression analysis using backward elimination feature selection method

Enter the following variables as input:

Sex, age, ho_incm, Dyslipidemia, DM_insulin, DM_duration, DM_po_med, Smoking, HE_sbp, HE_dbp, HE_wc, HE_BMI, HE_glu, HE_HbA1c, HE_chol, HE_TG, HE_ast, HE_alt, HE_HB, HE_BUN, HE_crea, HE_WBC, HE_Bplt, HE_Uacid

Please show the final prediction formula to predict E_DR

Draw the ROC curve of the E_DME prediction formula using the formula developed above. Show the optimal cutoff value

4. Organize formulas and input data

Please describe the formulas to predict E_DR (diabetic retinopathy) and E_DME (diabetic macular edema), which were developed above. Show the cutoff values, mentioned above, of each formula to identify the high-risk groups

What are the variables needed to calculate these formulas above?

5. Create HTML codes to build a risk calculator

Create a calculator that calculates the percentage risk scores of E_DR and E_DME using above formulas

Build the codes for the risk calculator written in html, css, and javascript in one html file

Design:

The text boxes must have rounded edges

Make this calculator look professional by creating a frame around it

Separate the title, input window and output window frames

Insert the title of the calculator in the title frame above the input window frame. The title is "Diabetic retinopathy (DR) and diabetic macular edema (DME) risk calculator"

Add the subtitle of the input window inside the frame: “Medical history and laboratory data”

Add the subtitle of the output window inside the frame: “DR and DME risk results”

The input frame should be above, and the output frame should be below. The title, input, and output frames must be aligned in order from the top

Input items:

Above-mentioned variables to calculate the formulas

Please show the units of the variables

Output items:

DR risk score (%)

Whether the calculation result corresponds to the DR high-risk or low-risk group

DME risk score (%)

Whether the calculation result corresponds to the DME high-risk or low-risk group

Set the size of the calculator to 800 by 800 pixels. Input items are in three columns. Please keep the inputs and calculation information as are

6. Validation of the developed DR and DME formulas

(After dragging the test dataset into the dialog window) This is the dataset for the external validation. Please draw ROC curves of the developed formulas above to predict DR and DME. Calculate the sensitivity and specificity at the optimal cutoff using the Youden's index