Benchmarking Machine Learning Algorithms on Blood Glucose Prediction for Type I Diabetes in Comparison With Classical Time-Series Models

Xie, Jinyu; Wang, Qian

doi:10.1109/tbme.2020.2975959

Cited by 102 publications

(58 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Although it is the simplest, its accuracy could be sufficient, and due to its simplicity, it is easily executable even by limited hardware. In any case, recent approximations using Least Absolute Shrinkage and Selection Operator (LASSO) regression have achieved acceptable accuracy and good performance [18].…”

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

“…Recently, Xie and Wang compared the performance of several black-box prediction algorithms on the OhioT1DM dataset [ 58 ]. In particular, they considered the Autoregressive model with exogenous inputs (ARX), elastic net, LASSO, Ridge and Huber regressions, gradient boosting trees, random forest, support vector regression (SVR) and two deep learning approaches: LSTM and Temporal Convolution Network (TCN).…”

Section: Glucose Predictionmentioning

confidence: 99%

Advanced Diabetes Management Using Artificial Intelligence and Continuous Glucose Monitoring Sensors

Vettoretti

Cappon

Facchinetti

et al. 2020

Sensors

View full text Add to dashboard Cite

Wearable continuous glucose monitoring (CGM) sensors are revolutionizing the treatment of type 1 diabetes (T1D). These sensors provide in real-time, every 1–5 min, the current blood glucose concentration and its rate-of-change, two key pieces of information for improving the determination of exogenous insulin administration and the prediction of forthcoming adverse events, such as hypo-/hyper-glycemia. The current research in diabetes technology is putting considerable effort into developing decision support systems for patient use, which automatically analyze the patient’s data collected by CGM sensors and other portable devices, as well as providing personalized recommendations about therapy adjustments to patients. Due to the large amount of data collected by patients with T1D and their variety, artificial intelligence (AI) techniques are increasingly being adopted in these decision support systems. In this paper, we review the state-of-the-art methodologies using AI and CGM sensors for decision support in advanced T1D management, including techniques for personalized insulin bolus calculation, adaptive tuning of bolus calculator parameters and glucose prediction.

show abstract

“…While an ideal classifier would obviously display a unitary AUC value, in general terms it can be stated that the higher the its value more robust is the decision made. The three proposed evaluation metrics are established and represent a standard for benchmarking DM methods [41], [42].…”

Section: Accmentioning

confidence: 99%

A Robust Decision-Making Framework Based on Collaborative Agents

et al. 2020

View full text Add to dashboard Cite

Making decisions under uncertainty is very challenging but necessary as most real-world scenarios are plagued by disturbances that can be generated internally, by the hardware itself, or externally, by the environment. Hence, we propose a general decision-making framework which can be adapted to optimally address the most heterogeneous real-world domains without being significantly affected by undesired disturbances. Our paper presents a multi-agent based structure in which agents are capable of individual decision-making but also interact to perform subsequent, and more robust, collaborative decisionmaking processes. The complexity of each software agent can be kept quite low without a deterioration of the performance since an intelligent and robust-to-uncertainty decision-making behaviour arises when their locally produced measure of support are shared and exploited collaboratively. We show that by equipping agents with classic computational intelligence techniques, to extract features and generate measure of supports, complex hybrid multi-agent software structures capable of handling uncertainty can be easily designed. The resulting multi-agent systems generated with this approach are based on a two-phases decision-making methodology which first runs parallel local decision making processes to then aggregate the corresponding outputs to improve upon the accuracy of the system. To highlight the potential of this approach, we provided multiple implementations of the general framework and compared them over four different application scenarios. Results are promising and show that having a second collaborative decisionmaking process is always beneficial.

show abstract

Benchmarking Machine Learning Algorithms on Blood Glucose Prediction for Type I Diabetes in Comparison With Classical Time-Series Models

Cited by 102 publications

References 22 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

Advanced Diabetes Management Using Artificial Intelligence and Continuous Glucose Monitoring Sensors

A Robust Decision-Making Framework Based on Collaborative Agents

Contact Info

Product

Resources

About