Predicting hypoglycemia in diabetic patients using data mining techniques

Eljil, Khouloud Safi; Qadah, Ghassan Z.; Pasquier, M.

doi:10.1109/innovations.2013.6544406

Cited by 18 publications

(24 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The reviewed articles had relied on both BG and other physiological (heart rate, ECG, skin impedance, and others) data, which of course involves different preprocessing strategies depending on the data type under consideration. Regarding the BG data, various preprocessing approaches had been used including differencing (derivative) BG values [27,28], CGM data reconstruction, or smoothing using different methods such as spline interpolation [29-33], a rough feature elimination, such as fast separability and correlation analysis algorithm [28,29], representing BG temporal change information [34], feature selection and feature ranking [35], filtering using Pearson’s correlation coefficient (PCC) and the t test, and the wrapper approach using greedy backward elimination [33]. The other physiological parameters (heart rate, ECG, skin impedance, and others) had been preprocessed using different methods such as normalization [36-38], feature extraction and selection [39,40], feature extraction using fast Fourier transform (FFT) [41], unsupervised restricted Boltzmann machine–based feature representation [42], filtering techniques such as Infinite impulse response high pass filter [41,43], correlation analysis [44-46], and transformation of frequency domain into time domain (FFT) [47].…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Data-Driven Blood Glucose Pattern Classification and Anomalies Detection: Machine-Learning Applications in Type 1 Diabetes

Woldaregay¹,

Årsand²,

Botsis³

et al. 2019

J Med Internet Res

114

View full text Add to dashboard Cite

Background Diabetes mellitus is a chronic metabolic disorder that results in abnormal blood glucose (BG) regulations. The BG level is preferably maintained close to normality through self-management practices, which involves actively tracking BG levels and taking proper actions including adjusting diet and insulin medications. BG anomalies could be defined as any undesirable reading because of either a precisely known reason (normal cause variation) or an unknown reason (special cause variation) to the patient. Recently, machine-learning applications have been widely introduced within diabetes research in general and BG anomaly detection in particular. However, irrespective of their expanding and increasing popularity, there is a lack of up-to-date reviews that materialize the current trends in modeling options and strategies for BG anomaly classification and detection in people with diabetes. Objective This review aimed to identify, assess, and analyze the state-of-the-art machine-learning strategies and their hybrid systems focusing on BG anomaly classification and detection including glycemic variability (GV), hyperglycemia, and hypoglycemia in type 1 diabetes within the context of personalized decision support systems and BG alarm events applications, which are important constituents for optimal diabetes self-management. Methods A rigorous literature search was conducted between September 1 and October 1, 2017, and October 15 and November 5, 2018, through various Web-based databases. Peer-reviewed journals and articles were considered. Information from the selected literature was extracted based on predefined categories, which were based on previous research and further elaborated through brainstorming. Results The initial results were vetted using the title, abstract, and keywords and retrieved 496 papers. After a thorough assessment and screening, 47 articles remained, which were critically analyzed. The interrater agreement was measured using a Cohen kappa test, and disagreements were resolved through discussion. The state-of-the-art classes of machine learning have been developed and tested up to the task and achieved promising performance including artificial neural network, support vector machine, decision tree, genetic algorithm, Gaussian process regression, Bayesian neural network, deep belief network, and others. Conclusions Despite the complexity of BG dynamics, there are many attempts to capture hypoglycemia and hyperglycemia incidences and the extent of an individual’s GV using different approaches. Recently, the advancement of diabetes technologies and continuous accumulation of self-collected health data have paved the way for popularity of machine learning in these tasks. According to the review, most of the identified studies used a theoretical threshold, which suffers from inter- and intrapatient variation. Therefore, future studies should consider the differenc...

show abstract

Section: Resultsmentioning

confidence: 99%

“…RF and DT have been mostly used in the context of hypoglycemia classification and detection tasks. For example, Eljil et al [27] proposed DTs using different techniques, namely, C4.5, J4.8, REPTree, bagging, and the cost-sensitive version of J4.8. Jung et al [74] also proposed DTs using new predictor variables using CGM data.…”

Section: Discussionmentioning

confidence: 99%

Data-Driven Blood Glucose Pattern Classification and Anomalies Detection: Machine-Learning Applications in Type 1 Diabetes

Woldaregay¹,

Årsand²,

Botsis³

et al. 2019

J Med Internet Res

114

View full text Add to dashboard Cite

show abstract

“…If its margin is largest then a hyperplane separating H is considered to be the top. The margin is the largest distance between two parallel hyperplanes to H on both sides that have no sample points between them [26,27]. It comes from the concept of risk minimization (the expected loss assessment function as a miss-classification of samples) that the higher the margin, the higher the generalization error of the classifier. )…”

Section: ) Preprocessingmentioning

confidence: 99%

“…ANN is used to model complex/nonlinear inter-relationships between inputs and outputs, for extracting significant patterns. In [26], ANN based classifier is used to model Diabetes dataset. The proposed ANN classifier has --configuration, where (the number of attributes to the model inputs), is the number of neurons in the hidden layer, where (using one hidden layer), and is the number of outputs that is equal one.…”

Section: ) Preprocessingmentioning

confidence: 99%

Prediction of Potential-Diabetic Obese-Patients using Machine Learning Techniques

Ali¹,

Elkadi²,

Safwat³

et al. 2019

IJACSA

View full text Add to dashboard Cite

Diabetes is a disease that is chronic. Improper blood glucose control may cause serious complications in diabetic patients as heart and kidney disease, strokes, and blindness. Obesity is considered to be a massive risk factor of type 2 diabetes. Machine Learning has been applied to many medical health aspects. In this paper, two machine learning techniques were applied; Support Vector Machine (SVM) and Artificial Neural Network (ANN) to predict diabetes mellitus. The proposed techniques were applied on a real dataset from Al-Kasr Al-Aini Hospital in Giza, Egypt. The models were examined using four-fold cross validation. The results were conducted from two phases in which forecasting patients with fatty liver disease using Support Vector Machine in the first phase reached the highest accuracy of 95% when applied on 8 attributes. Then, Artificial Neural Network technique to predict diabetic patients were applied on the output of phase 1 and another different 8 attributes to predict non-diabetic, pre-diabetic and diabetic patients with accuracy of 86.6%.

show abstract

“…Most of the previous work on predicting glucose by computational means focuses exclusively on the use of CGMS [25]. Because of the high CGMS measurement rate, prediction of future glucose may be more accurate than using BGM.…”

Section: Related Workmentioning

confidence: 99%

A Diabetes Management Information System with Glucose Prediction

Lélis¹,

Goulart

2018

Information

View full text Add to dashboard Cite

Diabetes has become a serious health concern. The use and popularization of blood glucose measurement devices have led to a tremendous increase on health for diabetics. Tracking and maintaining traceability between glucose measurements, insulin doses and carbohydrate intake can provide useful information to physicians, health professionals, and patients. This paper presents an information system, called GLUMIS (GLUcose Management Information System), aimed to support diabetes management activities. It is made of two modules, one for glucose prediction and one for data visualization and a reasoner to aid users in their treatment. Through integration with glucose measurement devices, it is possible to collect historical data on the treatment. In addition, the integration with a tool called the REALI System allows GLUMIS to also process data on insulin doses and eating habits. Quantitative and qualitative data were collected through an experimental case study involving 10 participants. It was able to demonstrate that the GLUMIS system is feasible. It was able to discover rules for predicting future values of blood glucose by processing the past history of measurements. Then, it presented reports that can help diabetics choose the amount of insulin they should take and the amount of carbohydrate they should consume during the day. Rules found by using one patient’s measurements were analyzed by a specialist that found three of them to be useful for improving the patient’s treatment. One such rule was “if glucose before breakfast ∈ [ 47 , 89 ] , then glucose at afternoon break in [ 160 , 306 ]”. The results obtained through the experimental study and other verifications associated with the algorithm created had a double objective. It was possible to show that participants, through a questionnaire, viewed the visualizations as easy, or very easy, to understand. The secondary objective showed that the innovative algorithm applied in the GLUMIS system allows the decision maker to have much more precision and less loss of information than in algorithms that require the data to be discretized.

show abstract

Predicting hypoglycemia in diabetic patients using data mining techniques

Cited by 18 publications

References 14 publications

Data-Driven Blood Glucose Pattern Classification and Anomalies Detection: Machine-Learning Applications in Type 1 Diabetes

Data-Driven Blood Glucose Pattern Classification and Anomalies Detection: Machine-Learning Applications in Type 1 Diabetes

Prediction of Potential-Diabetic Obese-Patients using Machine Learning Techniques

A Diabetes Management Information System with Glucose Prediction

Contact Info

Product

Resources

About