Tutorial

for the Computer Program to calculate the mean amplitude of glycemic excursions (MAGE)

This program can calculate the MAGE along with the mean sensor glucose (Mean SG), and the standard deviation (SD) from processed Continuous Glucose Monitoring System (CGMS) data. For this purpose, it is recommended to use sensor glucose values with 5-min updates collected over a maximum of 72 hours.

Fig. Program

Preferences

Menu: File - Preferences

Before you start the calculation, indicate in which measuring units the imported glucose data are provided and how the data and glucose curves should be displayed. You can choose between
mg/dL or mmol/l; 1 mmol/l is equal to 18 mg/dL.

Glucose Data import

The data structure of CGM data may differ, depending on the system that has been used for recording. This program version uses already processed data. Data must have the following format to be entered (txt-file with tab as separation sign). For example:

ICode_P-00121987_22-Oct-06
Sample Date Time Sensor Value
0 19-Oct-06 16:49
1 19-Oct-06 16:54
2 19-Oct-06 16:59
3 19-Oct-06 17:04
4 19-Oct-06 17:09
5 19-Oct-06 17:14
6 19-Oct-06 17:19
7 19-Oct-06 17:24
8 19-Oct-06 17:29
9 19-Oct-06 17:34
10 19-Oct-06 17:39
11 19-Oct-06 18:54 11.1
12 19-Oct-06 18:59 10.8
13 19-Oct-06 19:04 9.9
14 19-Oct-06 19:09 8.9
15 19-Oct-06 19:14 7.7
16 19-Oct-06 19:19 7.3
17 19-Oct-06 19:24 6.2
18 19-Oct-06 19:29 6.1
19 19-Oct-06 19:34 5.4
20 19-Oct-06 19:39 5.2

CGM data for several days of glucose recordings may be stored in this file. Define the time frame, i.e. duration of glucose recordings, according to your needs. However, data for at least 24-hour continuous measurements should be available. The time intervals of setups for the glucose measurements can be selected, e.g. 5-, 10-, 20-, 30-min intervals etc. Leaps in time or in date may lead to wrong results. The day the calculation starts, will be defined by the first glucose measurement.

First line of the file: A patient ID may be entered.
Window "Sensor Value": Entering no glucose values at all or even zero interprets the programm as missing value. The glucose curve will then linearly be continued between adjacent measuring points.

Menu: File - Import sensor data

Fig. window sensor data
Select starting date and interval of glucose updates in this window.

Clicking on button [Curves] allows preview of single days of glucose recordings.

Clicking on button [along] ] imports the data

The time interval between the glucose setups defines how the available values will be used. Values within the defined time interval will be used to calculate the MAGE. Smoothing of the curve may be achieved if larger time intervals, containing more than one glucose value (data points), will be considered.

Mean sensor glucose (Mean SG) and standard deviation (SD) will automatically be computed along with MAGE values, following import of the data.

Calculation

Prior to starting the calculation, define, according to the glucose profile data, whether the first and/or last glucose value of the time series ought to be considered as an extreme value (real minimum or maximum of the curve).

Fig. Start Button

Start calculation of the MAGE by clicking the button [Calculate].
As a result, a table will be displayed, showing the extremes and the time points when they appeared.

The MAGE Algorithm

The computation algorithm is based on successive approximation, using CGM datasets. The datasets provide glucose updates usually at 5-min intervals over 72-h periods. Calculation of MAGE by the software includes the following steps:

  1. Determination of all local maximums (max) and minimums (min).
    Note that only these values will be used for continued calculation along the time axis!
  2. Evaluation of the max-min pairs by the 1 SD criterion.
    If the difference max-min exeeds 1 SD, the value will be used in the ongoing calculation. If not, it will be disregarded. Note that only values are included if both edges of the glucose excursion exeed the value of 1 SD!
  3. A new evaluation cycle begins.

Example of glucose monitoring over 2 days with 15 min time intervals; SD = 75.2 mg/dL Fig. Example

This example shows that the difference (Δ) between the local maximum (7th data point) and the 1st minimum at 113 mg/dL exceeds 1 SD. The Δ of the corresponding local minimum is 63 mg/dL and is below the 1 SD level. Thus, both extremes will be disregarded. The following local maximum at 432 mg/dL satisfies the 1 SD criterion and will be included in the ongoing calculation. Evaluation subsequently proceeds accordingly until the end of the glucose recording.