A nonlinear blind identification approach to modeling of diabetic patients

Novara, Carlo; Pour, N. Mohammad; Vincent, Tyrone L.; Grassi, Giorgio

doi:10.3182/20140824-6-za-1003.01573

Cited by 8 publications

(10 citation statements)

References 29 publications

(20 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These various unmeasured disturbances such as food, physical activity, emotions and actuator errors are the main sources of the glucose fluctuating and cause the chattering phenomena in the insulin infusion rate. In this paper, a hybrid technique for BG [11] 0.000272 0.015499 hybrid control [27] 0.000036 0.003677 HoSM control [30] 0.004901 0.067432 HANFIS-PSO data fusion control 0.000034 0.003255 regulation by insulin injections is proposed to overcome these drawbacks.…”

Section: Discussionmentioning

confidence: 99%

“…The author thanks Iranian Diabetes Society for the strong technical support and research grant. [11] 0.0140 1.5897 hybrid control [27] 0.0178 1.5982 HoSM control [30] 0.0148 1.6328 HANFIS-PSO data fusion control 0.0122 1.5989 [11] 0.0413 3.1397 hybrid control [27] 0.0214 1.9659 HoSM control [30] 0.5049 8.3760 HANFIS-PSO data fusion control 0.0157 1.9244…”

Section: Acknowledgmentmentioning

confidence: 99%

See 1 more Smart Citation

Chattering ‐ free hybrid adaptive neuro‐fuzzy inference system‐particle swarm optimisation data fusion‐based BG‐level control

Karsaz

2020

IET Systems Biology

View full text Add to dashboard Cite

In this study, a closed-loop control scheme is proposed for the glucose-insulin regulatory system in type-1 diabetic mellitus (T1DM) patients. Some innovative hybrid glucose-insulin regulators have combined artificial intelligence such as fuzzy logic and genetic algorithm with well known Palumbo model to regulate the blood glucose (BG) level in T1DM patients. However, most of these approaches have focused on the glucose reference tracking, and the qualitative of this tracking such as chattering reduction of insulin injection has not been well-studied. Higher-order sliding mode (HoSM) controllers have been employed to attenuate the effect of chattering. Owing to the delayed nature and non-linear property of glucose-insulin mechanism as well as various unmeasurable disturbances, even the HoSM methods are partly successful. In this study, data fusion of adaptive neuro-fuzzy inference systems optimised by particle swarm optimisation has been presented. The excellent performance of the proposed hybrid controller, i.e. desired BG-level tracking and chattering reduction in the presence of daily glucose-level disturbances is verified.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Acknowledgmentmentioning

confidence: 99%

Chattering ‐ free hybrid adaptive neuro‐fuzzy inference system‐particle swarm optimisation data fusion‐based BG‐level control

Karsaz

2020

IET Systems Biology

View full text Add to dashboard Cite

show abstract

“…The main challenge of physiological models is the lack of generalization capability and need support from data for higher prediction performance. Data-driven approaches are mainly based on machine learning methods such as fuzzy logic and rule-based models [14], multi-modal approaches [15,16] autoregressive models [17,18], support vector machine [19] and artificial neural networks models [20]. The hybrid approach includes physiological models such as glucose digestion and absorption, insulin absorptions, exercise, and other events.…”

Section: Blood Glucose Prediction Researchmentioning

confidence: 99%

Blood Glucose Forecasting using LSTM Variants under the Context of Open Source Artificial Pancreas System

Wang¹,

Lewis³

2020

Proceedings of the Annual Hawaii International Conference on System Sciences

View full text Add to dashboard Cite

High accuracy of blood glucose prediction over the long term is essential for preventative diabetes management. The emerging closed-loop insulin delivery system such as the artificial pancreas system (APS) provides opportunities for improved glycaemic control for patients with type 1 diabetes. Existing blood glucose studies are proven effective only within 30 minutes but the accuracy deteriorates drastically when the prediction horizon increases to 45 minutes and 60 minutes. Deep learning, especially for long short term memory (LSTM) and its variants have recently been applied in various areas to achieve state-of-the-art results in tasks with complex time series data. In this study, we present deep LSTM based models that are capable of forecasting long term blood glucose levels with improved prediction and clinical accuracy. We evaluate our approach using 20 cases(878,000 glucose values) from Open Source Artificial Pancreas System (OpenAPS). On 30-minutes and 45-minutes prediction, our Stacked-LSTM achieved the best performance with Root-Mean-Square-Error (RMSE) marks 11.96 & 15.81 and Clark-Grid-ZoneA marks 0.887 & 0.784. In terms of 60-minutes prediction, our ConvLSTM has the best performance with RMSE = 19.6 and Clark-Grid-ZoneA=0.714. Our models outperform existing methods in both prediction and clinical accuracy. This research can hopefully support patients with type 1 diabetes to better manage their behavior in a more preventative way and can be used in future real APS context.

show abstract

“…In [16], an algorithm was developed that uses the difference between predictions from a simple glucose-insulin model and CGM measurements to detect meal occurrences and to concurrently estimate u ra . A recent study in [17] phrased the problem of identifying glucose fluxes as a blind identification problem for identifying both model parameters and unmeasured disturbances simultaneously from CGM and insulin pump measurements.…”

Section: Introductionmentioning

confidence: 99%

Sparse Reconstruction of Glucose Fluxes Using Continuous Glucose Monitors

Al-Matouq

Laleg‐Kirati

Novara

et al. 2020

IEEE/ACM Trans. Comput. Biol. and Bioinf.

View full text Add to dashboard Cite

A new technique for estimating postprandial glucose flux profiles without the use of glucose tracers is proposed. The technique assumes knowledge of patient parameters relevant to the glucose, insulin and endogoneous glucose production subsystems. A convex Lasso formulation is used to estimate the glucose fluxes that combines (1) the known patient parameters; (2) a sparse vector space encoding the space of plausible glucose flux profiles; (3) continuous glucose monitor measurements taken during the meal; (4) amount of insulin injected; (5) amount of meal carbohydrates and (6) an estimate of the initial conditions. Three glucose fluxes are estimated: glucose rate of appearance from the intestine; endogenous glucose production from the liver; insulin dependent glucose utilization and other important state variables. Sparse encoding of a large set of simulated glucose fluxes using the UVa Padova simulator is used so that a sparse representation of the space of plausible glucose flux profiles is obtained. The estimation technique was validated in both simulation and experiments on 3 T1DM patients undergoing the triple tracer meal protocol. The results indicate that the technique is capable of reproducing the triple tracer measurements while estimating the missing measurements for a certain model parameter selection.

show abstract

A nonlinear blind identification approach to modeling of diabetic patients

Cited by 8 publications

References 29 publications

Chattering ‐ free hybrid adaptive neuro‐fuzzy inference system‐particle swarm optimisation data fusion‐based BG‐level control

Chattering ‐ free hybrid adaptive neuro‐fuzzy inference system‐particle swarm optimisation data fusion‐based BG‐level control

Blood Glucose Forecasting using LSTM Variants under the Context of Open Source Artificial Pancreas System

Sparse Reconstruction of Glucose Fluxes Using Continuous Glucose Monitors

Contact Info

Product

Resources

About