25.01.2013 Views

Supplementary Data - Diabetes

Supplementary Data - Diabetes

Supplementary Data - Diabetes

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

SUPPLEMENTARY DATA<br />

where I is plasma insulin concetration, n is the fractional insulin clearance rate, VI is insulin volume of<br />

distribution and IDR is insulin delivery rate, i.e. the rate of appearance of insulin in plasma after hepatic<br />

extraction (HE). IDR, ISR and HE are linked by the following relationship:<br />

ISR(<br />

t)<br />

� IDR(<br />

t)<br />

HE(<br />

t)<br />

�<br />

ISR(<br />

t)<br />

thus<br />

ISR t � HE t �<br />

I� ( ) � 1 � ( )<br />

( t)<br />

� �n<br />

� I(<br />

t)<br />

�<br />

I(<br />

0)<br />

� I<br />

V<br />

I<br />

According to Campioni et al 2009, HE is described with a piecewise linear function:<br />

� HE<br />

�HEi�1<br />

�<br />

HE(<br />

t)<br />

� � t<br />

�<br />

�HE0<br />

� HEb<br />

i<br />

i<br />

� HE<br />

� t<br />

i�1<br />

i�1<br />

�<br />

�t�t� i�1<br />

t<br />

i�1<br />

� t � t<br />

(A11)<br />

©2012 American <strong>Diabetes</strong> Association. Published online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db11-1478/-/DC1<br />

i<br />

b<br />

i �1,<br />

2,...<br />

6<br />

wehere HEb is the basal hepatic extraction, that can be derived from basal conncetrations:<br />

HE<br />

b<br />

k<br />

�<br />

01<br />

� Cb<br />

�Vc<br />

� k 01 � I<br />

k � C �V<br />

01<br />

b<br />

b<br />

�V<br />

I<br />

(A12)<br />

(A13)<br />

(A14)<br />

A table containing the definition of all the minimal model indices can be found in [Cobelli et al, Am. J.<br />

Physiol 2007], Table 2(3).<br />

Parameter Estimation<br />

Oral Glucose Minimal Model was numerically identified by nonlinear least squares, implemented in<br />

Matlab ® R2010b. Measurement error on glucose data was assumed to be independent, gaussian, with<br />

zero mean and known standard deviation (CV = 2%). Insulin concentration is the model forcing function<br />

and was assumed to be known without error. Model identification requires a number of assumptions<br />

which were discussed in detail in [Dalla Man et al, IEEE Trans Biomed Eng 2002] (4). Briefly, to ensure<br />

its a priori identifiability one has to assume values for V and SG. Here we fixed them to the median<br />

obtained with a triple tracer method in [Dalla Man et al Am J. Physiol, 2004] (5), i.e. V=1.45 dl/kg,<br />

SG=0.025 min -1 (median was preferred to mean values since parameters are not normally distributed).<br />

To improve numerical identifiability of the remaining parameters p2, p3, �i (i=1..7) a gaussian bayesian<br />

prior was considered on the square root of p2 ref (SQRp2 ref ), which is normally distributed: SQRp2 =0.11<br />

min -1/2 , SD=10%. Finally, a constraint was imposed to guarantee that the area under the estimated Rameal<br />

equals the total amount of ingested glucose, D, multiplied by the fraction of the ingested dose that is<br />

actually absorbed, f (fixed to the median of reference values f=0.9): this constraint provides an<br />

additional relationship among the unknown parameters �i, thus reducing the number of unknowns by<br />

one.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!