A multi-level hypoglycemia early alarm system based on sequence pattern mining

Xia Yu,Yawen Zhang,Hongru Li,Tao Yang,Yiming Li,Qing Miao,Yehong Yang,Ning Ma,Junjun Tao

doi:10.1186/s12911-021-01389-x

Xia Yu, Yawen Zhang + Show 7 more

Open Access

https://doi.org/10.1186/s12911-021-01389-x

Copy DOI

Abstract

BackgroundEarly alarm of hypoglycemia, detection of asymptomatic hypoglycemia, and effective control of blood glucose fluctuation make a great contribution to diabetic treatment. In this study, we designed a multi-level hypoglycemia early alarm system to mine potential information in Continuous Glucose Monitoring (CGM) time series and improve the overall alarm performance for different clinical situations.MethodsThrough symbolizing the historical CGM records, the Prefix Span was adopted to obtain the early alarm/non-alarm frequent sequence libraries of hypoglycemia events. The longest common subsequence was used to remove the common frequent sequence for achieving the hypoglycemia early alarm in different clinical situations. Then, the frequent sequence pattern libraries with different risk thresholds were designed as the core module of the proposed multi-level hypoglycemia early alarm system.ResultsThe model was able to predict hypoglycemia events in the clinical dataset of level-I (sensitivity 85.90%, false-positive 23.86%, miss alarm rate 14.10%, average early alarm time 20.61 min), level-II (sensitivity 80.36%, false-positive 17.37%, miss alarm rate 19.63%, average early alarm time 27.66 min), and level-III (sensitivity 78.07%, false-positive 13.59%, miss alarm rate 21.93%, average early alarm time 33.80 min), respectively.ConclusionsThe proposed approach could effectively predict hypoglycemia events based on different risk thresholds to meet different prevention and treatment requirements. Moreover, the experimental results confirm the practicality and prospects of the proposed early alarm system, which reflects further significance in personalized medicine for hypoglycemia prevention.

Highlights

IntroductionDetection of asymptomatic hypoglycemia, and effective control of blood glucose fluctuation make a great contribution to diabetic treatment
Alarm of hypoglycemia, detection of asymptomatic hypoglycemia, and effective control of blood glucose fluctuation make a great contribution to diabetic treatment
As time series expressed in real numbers do not allow instant expression and extraction of patterns, most studies on sequential pattern mining are based on symbolic patterns

Summary

Introduction

Detection of asymptomatic hypoglycemia, and effective control of blood glucose fluctuation make a great contribution to diabetic treatment. One of the applications is designing hypoglycemia early alarm systems, which is an effective way to predict hypoglycemia events for diabetic patients. In 2009, Buckingham [5] used linear projection and statistical prediction alarm methods to predict hypoglycemia events and provided a timely response (shutting down the insulin pump for 90 min). In 2013, Bayrak et al [7] used recursive autoregressive partial least squares (RARPLS) algorithm to model the CGM sensor data and predict the future glucose concentration for the hypoglycemia alarm system. In 2020, Vehí [10] designed a hybrid model with four machine learning algorithms to solve the safety problems of diabetes management, including grammar evolution, support vector machines, artificial neural network, and data mining. For the glucose prediction model, it is unavoidable that the prediction value generally lags behind the real value, and the accuracy greatly affects the early alarm performance of hypoglycemia

Methods

Results

Discussion

Conclusion