Multiresolution Wavelet Analysis of ECG Features for Blood Glucose Prediction and Study of a Potential Market Entry for an ECG-Blood Glucose Monitoring Device

Background: Nowadays, the tools and methods to measure Blood Glucose Levels (BGLs) are quite, invasive, and inconvenient, especially for diabetes patients. For these reasons, several studies are currently focusing on predicting and measuring the glycaemia using instruments such as near infrared spectroscopy (NIR) and Raman spectroscopy. Recently, a few studies tried to use the electrocardiogram (ECG) and its multiple features to establish a link between them and the blood glucose levels in order to predict the glycaemia without the need to prick the skin. The techniques used rely mostly on the extraction of a few segments and intervals from the ECG signals of patients, that have in parallel, their BGLs monitored continuously to establish a linear regression between some of these ECG parameters and the glycaemia.
Aims: This research relies on data (ECG and BGLs) obtained from Singular Wings Medical (SWM) company. The purpose is to use the multiresolution wavelet approach to extract heart rate (HR) parameters-related information of the ECG signals and link them to BGLs to analyse in which aspects the Blood Glucose (BG) and ECG are linked together. Research around the ECG and its link with BG could have an impact on the thought and design of a potential device that could measure glycemia from one or more of its features.
Material and methods: Two studies focused on following and analysing the HR and BG variations per days for several candidates, for which the information have been provided by SWM. The BG data, measured by the Freestyle Libre 2, were sampled at 1 data every 15 min and were directly accessible with excel files provided by SWM. The ECG sampled at 250 Hz and as 10 seconds signals were provided as srj files and processed under MATLAB software. The HR information was retrieved from 2 methods: using Wavelet transform (WT) with MATLAB software and using SWM calculations as a comparison and validation of the HR obtained from the WT.
The third study focused on correlation tests, with Thonny software (Python), between statistical parameters (mean, median, mode, standard deviation and variance) of BG and HR calculated previously with MATLAB.
To go further, the possibility of putting an ECG-BG device on the french market has been analysed through different tools, essentially to collect diabetic population needs and bring out recommendations on how to enter the market. This has been done through a questionnaire, a benchmarking and a swot analysis.
Results: Results from two first study showed a trend of the HR variations following the ones from BG. That’s to say when the BG increases, the HR tends to rise too and vice-versa. Correlation tests also showed that the median BG and mean HR, median BG and mode HR and mode BG and mode HR have positive linear relationship between each other, with respective correlation coefficients of 0,75998, 0,72 and 0,7122.
The market analysis highlighted the major key players (Abbott, Medtronic and Dexcom) in the Blood Glucose Monitoring (BGM) devices market as well as the strong points making their devices widely used among others. Also, the questionnaire has been filled by 155 diabetic persons that have high hopes for improvements of future non-invasive devices, especially on the accuracy, and sensor characteristics (size and thickness).
Conclusion: The BG seems related to the HR according to the results of this study: the more the BG, the higher the HR. Still more investigations need to be done on these conclusions, especially first by knowing the profile of the persons with which the data are processed, so that some external parameters that could have interfered and influenced the HR can be excluded, enabling the observations of the conclusions established previously to be fully interpreted and conclusive.
Furthermore, the market analysis highlighted several possible improvements on the diabetes devices market, especially the non-invasive one that is still new and full of opportunities.