Eleni I. Georga

Multivariate Prediction of Subcutaneous Glucose Concentration in Type 1 Diabetes Patients Based on Support Vector Regression

Data-driven techniques have recently drawn significant interest in the predictive modeling of subcutaneous (s.c.) glucose concentration in type 1 diabetes. In this study, the s.c. glucose prediction is treated as a multivariate regression problem, which is addressed using support vector regression (SVR). The proposed method is based on variables concerning: 1) the s.c. glucose profile; 2) the plasma insulin concentration; 3) the appearance of meal-derived glucose in the systemic circulation; and 4) the energy expenditure during physical activities. Six cases corresponding to different combinations of the aforementioned variables are used to investigate the influence of the input on the daily glucose prediction. The proposed method is evaluated using a dataset of 27 patients in free-living conditions. Tenfold cross validation is applied to each dataset individually to both optimize and test the SVR model. In the case, where all the input variables are considered, the average prediction errors are 5.21, 6.03, 7.14, and 7.62 mg/dl for 15-, 30-, 60-, and 120-min prediction horizons, respectively. The results clearly indicate that the availability of multivariable data and their effective combination can significantly increase the accuracy of both short-term and long-term predictions.

Evaluation of short-term predictors of glucose concentration in type 1 diabetes combining feature ranking with regression models

Glucose concentration in type 1 diabetes is a function of biological and environmental factors which present high inter-patient variability. The objective of this study is to evaluate a number of features, which are extracted from medical and lifestyle self-monitoring data, with respect to their ability to predict the short-term subcutaneous (s.c.) glucose concentration of an individual. Random forests (RF) and RReliefF algorithms are first employed to rank the candidate feature set. Then, a forward selection procedure follows to build a glucose predictive model, where features are sequentially added to it in decreasing order of importance. Predictions are performed using support vector regression or Gaussian processes. The proposed method is validated on a dataset of 15 type diabetics in real-life conditions. The s.c. glucose profile along with time of the day and plasma insulin concentration are systematically highly ranked, while the effect of food intake and physical activity varies considerably among patients. Moreover, the average prediction error converges in less than d/2 iterations (d is the number of features). Our results suggest that RF and RReliefF can find the most informative features and can be successfully used to customize the input of glucose models.

An architecture for designing Future Internet (FI) applications in sensitive domains: Expressing the software to data paradigm by utilizing hybrid cloud technology

The emergency of cloud computing and Generic Enablers (GEs) as the building blocks of Future Internet (FI) applications highlights new requirements in the area of cloud services. Though, due to the current restrictions of various certification standards related with privacy and safety of health related data, the utilization of cloud computing in such area has been in many instances unlawful. Here, we focus on demonstrating a “software to data” provisioning solution to propose a mapping of FI application use case requirements to software specifications (using GEs). The aim is to establish a provider to consumer cloud setting wherein no sensitive data will be exchanged but it will reside at the back-end site. We propose a prototype architecture that covers the cloud management layer and the operational features that manage data and Internet of Things devices. To show a real life scenario, we present the use case of the diabetes care and a FI application that includes various GEs.

A Glucose Model Based on Support Vector Regression for the Prediction of Hypoglycemic Events Under Free-Living Conditions

BACKGROUND The prevention of hypoglycemic events is of paramount importance in the daily management of insulin-treated diabetes. The use of short-term prediction algorithms of the subcutaneous (s.c.) glucose concentration may contribute significantly toward this direction. The literature suggests that, although the recent glucose profile is a prominent predictor of hypoglycemia, the overall patients context greatly impacts its accurate estimation. The objective of this study is to evaluate the performance of a support vector for regression (SVR) s.c. glucose method on hypoglycemia prediction. MATERIALS AND METHODS We extend our SVR model to predict separately the nocturnal events during sleep and the non-nocturnal (i.e., diurnal) ones over 30-min and 60-min horizons using information on recent glucose profile, meals, insulin intake, and physical activities for a hypoglycemic threshold of 70 mg/dL. We also introduce herein additional variables accounting for recurrent nocturnal hypoglycemia due to antecedent hypoglycemia, exercise, and sleep. SVR predictions are compared with those from two other machine learning techniques. RESULTS The method is assessed on a dataset of 15 patients with type 1 diabetes under free-living conditions. Nocturnal hypoglycemic events are predicted with 94% sensitivity for both horizons and with time lags of 5.43 min and 4.57 min, respectively. As concerns the diurnal events, when physical activities are not considered, the sensitivity is 92% and 96% for a 30-min and 60-min horizon, respectively, with both time lags being less than 5 min. However, when such information is introduced, the diurnal sensitivity decreases by 8% and 3%, respectively. Both nocturnal and diurnal predictions show a high (>90%) precision. CONCLUSIONS Results suggest that hypoglycemia prediction using SVR can be accurate and performs better in most diurnal and nocturnal cases compared with other techniques. It is advised that the problem of hypoglycemia prediction should be handled differently for nocturnal and diurnal periods as regards input variables and interpretation of results.

Glucose Prediction in Type 1 and Type 2 Diabetic Patients Using Data Driven Techniques

Diabetes mellitus, commonly referred to as diabetes, is a group of metabolic diseases characterized by high blood glucose concentrations resulting from defects in insulin secretion, insulin action or both [American Diabetes Association, 2008a]. Diabetes has been classified into two major categories, namely, type 1 and type 2 diabetes. Type 1 diabetes, which accounts for only 5-10% of those with diabetes, is caused by the cell-mediated autoimmune destruction of the insulin producing β-cells in the pancreas leading to absolute insulin deficiency. On the other hand, type 2 diabetes is a more prevalent category (i.e. accounts for ~90-95% of those with diabetes) and is a combination of resistance to insulin action and an inadequate compensatory insulin secretion. The chronic hypergycemia of diabetes is associated with long-term microvascular (diabetic neuropathy, nephropathy and retinopathy) and macrovascular (coronary artery disease, peripheral arterial disease, and stroke) complications. Diabetes treatment requires the control of clinical and non-clinical variables affecting the blood glucose metabolism [American Diabetes Association, 2008b]. It is widely recognized that the tight glycemic control can prevent or reduce the progress of many long-term complications of diabetes. However, a major limiting factor in the glycemic management of type 1 and insulin treated type 2 diabetes is hypoglycemia, which is the condition where the blood glucose is much lower than normal levels. Thus, for most patients with type 1 diabetes, either using multiple insulin injections or insulin pump therapy, self-monitoring of blood glucose should be carried out three or more times a day, whereas, for patients using less frequent insulin injections or non-insulin therapies, the self-monitoring of blood glucose could be useful in achieving their glycemic targets. Recently, continuous glucose monitoring (CGM) systems have been developed which provide many significant benefits in diabetes management, especially for those patients with hypoglycaemia unawareness. Moreover, diabetes control further necessitates the monitoring and analysis of patient’s contextual information, such as medication, diet, physical activity and his overall lifestyle. For instance, in type 1 diabetic patients, exercise can cause hypoglycemia in the case where the medication dose or the carbohydrate consumption is not altered. In addition to the general guidelines that the patient follows during his daily life, several diabetes management systems have been proposed to further assist the patient in the selfmanagement of the disease. One of the most essential components of a diabetes

A predictive model of subcutaneous glucose concentration in type 1 diabetes based on Random Forests

In this study, an individualized predictive model of the subcutaneous glucose concentration in type 1 diabetes is presented, which relies on the Random Forests regression technique. A multivariate dataset is utilized concerning the s.c. glucose profile, the plasma insulin concentration, the intestinal absorption of meal-derived glucose and the daily energy expenditure. In an attempt to capture daily rhythms in glucose metabolism, we also introduce a time feature in the predictive analysis. The dataset comes from the continuous multi-day recordings of 27 type 1 patients in free-living conditions. Evaluating the performance of the proposed method by 10-fold cross validation, an average RMSE of 6.60, 8.15, 9.25 and 10.83 mg/dl for 15, 30, 60 and 120 min prediction horizons, respectively, was attained.

Short-term vs. long-term analysis of diabetes data: Application of machine learning and data mining techniques

Chronic care of diabetes comes with large amounts of data concerning the self- and clinical management of the disease. In this paper, we propose to treat that information from two different perspectives. Firstly, a predictive model of short-term glucose homeostasis relying on machine learning is presented with the aim of preventing hypoglycemic events and prolonged hyperglycemia on a daily basis. Second, data mining approaches are proposed as a tool for explaining and predicting the long-term glucose control and the incidence of diabetic complications.

Prediction of glucose concentration in type 1 diabetic patients using support vector regression

Diabetic patients must adhere continually to a complex daily regime in order to maintain the blood glucose levels within a safe range. Many factors impact glucose variations such as diet, medication and exercise. This work presents a modeling methodology for glucose prediction in type 1 diabetic patients. The physiological processes related to diabetes (i.e. insulin absorption, gut absorption) as well as the effects of exercise on blood glucose and insulin dynamics are quantified using compartmental models. Furthermore, the method employs Support Vector Machines for Regression to provide predictions of glucose concentration. The predictive capabilities of the resulting model are evaluated using data from three type 1 diabetic patients. The Clarkes Error Grid Analysis is used to assess the clinical utility of the proposed prediction method.

Non-linear dynamic modeling of glucose in type 1 diabetes with kernel adaptive filters

We propose a non-linear recursive solution to the problem of short-term prediction of glucose in type 1 diabetes. The Fixed Budget Quantized Kernel Least Mean Square (QKLMS-FB) algorithm is employed to construct a univariate model of subcutaneous glucose concentration, which: (i) handles nonlinearities by transforming the input space into a high-dimensional Reproducing Kernel Hilbert Space and, (ii) finds a sparse solution by retaining a representative subset of the training input vectors. The dataset comes from the continuous multi-day recordings of 15 type 1 patients in free-living conditions. QKLMS-FB produces an average root mean squared error of 18.66±3.19 mg/dl for a prediction horizon of 30 min with 82.04% of hypoglycemic readings and 93.30% of hyperglycemic ones being classified as clinically accurate or with benign errors. The effect of the prediction horizon is more evident in the hypoglycemic range.We propose a non-linear recursive solution to the problem of short-term prediction of glucose in type 1 diabetes. The Fixed Budget Quantized Kernel Least Mean Square (QKLMS-FB) algorithm is employed to construct a univariate model of subcutaneous glucose concentration, which: (i) handles nonlinearities by transforming the input space into a high-dimensional Reproducing Kernel Hilbert Space and, (ii) finds a sparse solution by retaining a representative subset of the training input vectors. The dataset comes from the continuous multi-day recordings of 15 type 1 patients in free-living conditions. QKLMS-FB produces an average root mean squared error of 18.66±3.19 mg/dl for a prediction horizon of 30 min with 82.04% of hypoglycemic readings and 93.30% of hyperglycemic ones being classified as clinically accurate or with benign errors. The effect of the prediction horizon is more evident in the hypoglycemic range.

Online prediction of glucose concentration in type 1 diabetes using extreme learning machines.

We propose an online machine-learning solution to the problem of nonlinear glucose time series prediction in type 1 diabetes. Recently, extreme learning machine (ELM) has been proposed for training single hidden layer feed-forward neural networks. The high accuracy and fast learning speed of ELM drive us to investigate its applicability to the glucose prediction problem. Given that diabetes self-monitoring data are received sequentially, we focus on online sequential ELM (OS-ELM) and online sequential ELM kernels (KOS-ELM). A multivariate feature set is utilized concerning subcutaneous glucose, insulin therapy, carbohydrates intake and physical activity. The dataset comes from the continuous multi-day recordings of 15 type 1 patients in free-living conditions. Assuming stationarity and evaluating the performance of the proposed method by 10-fold cross- validation, KOS-ELM were found to perform better than OS-ELM in terms of prediction error, temporal gain and regularity of predictions for a 30-min prediction horizon.