Gaussian process-based model predictive control of blood glucose for patients with type 1 diabetes mellitus

Ortmann, Lukas; Shi, Dawei; Dassau, Eyal; Doyle, Francis J.; Leonhardt, Steffen; Misgeld, Berno J.E.

doi:10.1109/ascc.2017.8287323

Cited by 17 publications

(11 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are other methods using the same sort of techniques, for example, Gaussian Processes (GP) with Radial Basis Function Kernels (RBF) [19], which permit overall uniformity and limitless levels of basic functions. However, these are not often employed, although some researchers have used such techniques and the results have shown promise [20]. Some recent research has also looked at GP [21], examining the potential for automatic insulin delivery that could reduce the number of hypoglycemic events.…”

Section: Related Workmentioning

confidence: 99%

A Comparison of Feature Selection and Forecasting Machine Learning Algorithms for Predicting Glycaemia in Type 1 Diabetes Mellitus

et al. 2021

View full text Add to dashboard Cite

Type 1 diabetes mellitus (DM1) is a metabolic disease derived from falls in pancreatic insulin production resulting in chronic hyperglycemia. DM1 subjects usually have to undertake a number of assessments of blood glucose levels every day, employing capillary glucometers for the monitoring of blood glucose dynamics. In recent years, advances in technology have allowed for the creation of revolutionary biosensors and continuous glucose monitoring (CGM) techniques. This has enabled the monitoring of a subject’s blood glucose level in real time. On the other hand, few attempts have been made to apply machine learning techniques to predicting glycaemia levels, but dealing with a database containing such a high level of variables is problematic. In this sense, to the best of the authors’ knowledge, the issues of proper feature selection (FS)—the stage before applying predictive algorithms—have not been subject to in-depth discussion and comparison in past research when it comes to forecasting glycaemia. Therefore, in order to assess how a proper FS stage could improve the accuracy of the glycaemia forecasted, this work has developed six FS techniques alongside four predictive algorithms, applying them to a full dataset of biomedical features related to glycaemia. These were harvested through a wide-ranging passive monitoring process involving 25 patients with DM1 in practical real-life scenarios. From the obtained results, we affirm that Random Forest (RF) as both predictive algorithm and FS strategy offers the best average performance (Root Median Square Error, RMSE = 18.54 mg/dL) throughout the 12 considered predictive horizons (up to 60 min in steps of 5 min), showing Support Vector Machines (SVM) to have the best accuracy as a forecasting algorithm when considering, in turn, the average of the six FS techniques applied (RMSE = 20.58 mg/dL).

show abstract

Section: Related Workmentioning

confidence: 99%

A Comparison of Feature Selection and Forecasting Machine Learning Algorithms for Predicting Glycaemia in Type 1 Diabetes Mellitus

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The training data for the Gaussian Process consists of these data points and there corresponding time stamps. For more details, see [17].…”

Section: A Training Data Calculationmentioning

confidence: 99%

“…In this paper, we present an extension of our work on incorporating information about the changing insulin sensitivity into a controller for the insulin-glucose metabolism [17]. In contrast to previously published results on including the insulin sensitivity, we determine the effect of the changing insulin sensitivity during closed-loop.…”

Section: Introductionmentioning

confidence: 99%

Automated Insulin Delivery for Type 1 Diabetes Mellitus Patients using Gaussian Process-based Model Predictive Control

Ortmann

Shi

Dassau

et al. 2019

2019 American Control Conference (ACC)

Self Cite

View full text Add to dashboard Cite

The human insulin-glucose metabolism is a timevarying process, which is partly caused by the changing insulin sensitivity of the body. This insulin sensitivity follows a circadian rhythm and its effects should be anticipated by any automated insulin delivery system. This paper presents an extension of our previous work on automated insulin delivery by developing a controller suitable for humans with Type 1 Diabetes Mellitus. Furthermore, we enhance the controller with a new kernel function for the Gaussian Process and deal with noisy measurements, as well as, the noisy training data for the Gaussian Process, arising therefrom. This enables us to move the proposed control algorithm, a combination of Model Predictive Controller and a Gaussian Process, closer towards clinical application. Simulation results on the University of Virginia/Padova FDA-accepted metabolic simulator are presented for a meal schedule with random carbohydrate sizes and random times of carbohydrate uptake to show the performance of the proposed control scheme.

show abstract

“…The algorithm from Ortmann et al [17], is based on MPC and considers periodic insulin sensitivity (IS), which changes with a patient's circadian rhythm, to improve BG control. They use machine learning (ML) to predict the effect of the IS.…”

Section: Closing the Loop With Algorithmsmentioning

confidence: 99%

Towards an Artificial Pancreas: Software Architectural Model and Implementation for Personalized Insulin Administration

Jensen¹,

Juric

McClenaghan

et al. 2019

Proceedings of the Annual Hawaii International Conference on System Sciences

View full text Add to dashboard Cite

Research on an Artificial Pancreas has gained its momentum and focused on the processing of clinical data for continuous insulin administration. However, the overall research is rather sketchy, fragmented and not very well coordinated. In this paper, we propose an architectural model for creating software intensive environments, which address deficiencies of current solutions for insulin infusion. A new way of defining which data should be collected and which types of computations should be performed with the data is important if we wish to come close to the functioning of a natural pancreas. An excerpt of the proposed software architecture has been deployed using Watson Analytics and performed upon a selection of data collected from sensors, individual patient's input and persistent patient records.

show abstract

Gaussian process-based model predictive control of blood glucose for patients with type 1 diabetes mellitus

Cited by 17 publications

References 24 publications

A Comparison of Feature Selection and Forecasting Machine Learning Algorithms for Predicting Glycaemia in Type 1 Diabetes Mellitus

A Comparison of Feature Selection and Forecasting Machine Learning Algorithms for Predicting Glycaemia in Type 1 Diabetes Mellitus

Automated Insulin Delivery for Type 1 Diabetes Mellitus Patients using Gaussian Process-based Model Predictive Control

Towards an Artificial Pancreas: Software Architectural Model and Implementation for Personalized Insulin Administration

Contact Info

Product

Resources

About