Pavlo Tkachenko, Florian Reiterer, Luigi Del Re,
"Identification of T2D patient typologies susceptible to artificial pancreas usage"
: Final AMMODIT Conference "Mathematics for Life Sciences", 3-2019
Original Titel:
Identification of T2D patient typologies susceptible to artificial pancreas usage
Sprache des Titels:
Englisch
Original Buchtitel:
Final AMMODIT Conference "Mathematics for Life Sciences"
Original Kurzfassung:
In this work we are intended to explore which subgroup of patients is expected to benefit substantially from a potential artificial pancreas (AP) system and whether this subgroup can be identified using patient information that is a priori available (i.e. before the implementation of the AP on a patient). The key fact behind it is that AP is not expected to be necessary or even beneficial for every type 2 diabetes (T2D) patient, but only for a subgroup of insulin treated T2D patients that does not succeed in accomplishing the therapy goals with simpler insulin dosing approaches.
Against this background, a sample of insulin treated T2D patients was used to estimate the possible benefit of automated insulin dosing via an AP and compare it to the achievable performance with continuous subcutaneous insulin infusion (CSII) therapy. The analysis was done using state-of-the-art deviation analysis in an in silico testing environment.
The key result is that a simple hybrid AP approach would be beneficial for the vast majority of analyzed patients. However, for 60% of the analyzed patients improved settings of basal insulin dosing via CSII together with fixed bolus quantities is suficient for a good glycemic control as well. Most of the remaining patients exhibit significantly higher glycemic variability and diurnal variations in insulin needs and therefore require a more complex therapy scheme. For those patients AP seems to be the therapy of choice.
This subgroup could be identified a priori using only a limited number of features. Additionally, a continuous score that estimates the therapeutic outcomes of the different treatment options has been proposed and a regression analysis can be used to estimate the potential benefits a priori as well.