Open AccessDevelopment of non-invasive alcohol analyzer using Photoplethysmographytle
Author(s) -
Pornnapa Sanguansri,
Nattapat Apiwong-ngam,
Athipong Ngamjarurojana,
Supab Choopun
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2145/1/012059
Subject(s) - photoplethysmogram , spectrum analyzer , signal (programming language) , computer science , pattern recognition (psychology) , statistic , artificial intelligence , waveform , set (abstract data type) , statistics , mathematics , biomedical engineering , medicine , computer vision , telecommunications , radar , filter (signal processing) , programming language
Photoplethysmography (PPG) is one of the optical signals commonly used in clinical research for measuring the vital signs. Previously, PPG is often used as an indicator for detecting blood volume changes in the micro-vascular. The advantages of PPG signal mentioned in studies are non-invasive, lower operation cost, and the simplicity of the procedure. Although some the components of the PPG signal are not fully understood, it is generally accepted that it can provide valuable information in clinical study. Thus, it is interesting for finding a relation between PPG signal and blood alcohol concentration. The objective of this study is to classify two groups of ten-volunteer: (1) group of people who consumed alcohol and (2) non-consumed alcohol, by using the difference of PPG signals in these two groups and compared the results with fuel-cell breath alcohol analyzer. A set of PPG reflection data is recorded from optical sensors including the wavelength light of the red light and the infrared light from the fingertip of individuals. In additional, the changes of each signals for distinguishing two groups of volunteers are examined. The set of data is computed and analysed to find the correlation coefficient between significant variables in statistic domain. The analysis techniques are included (1) slope of the signals over time, (2) peak to peak of the heart rate, and (3) deep of waveform valley after rotation for training generalized linear (GLM) classifiers to create classification models. The accuracy of GLM classification can be obtained up to 87.50%. This suggests that PPG technique with our lab prototype has a potential for screening test to classify people who consumed alcohol and non-consumed alcohol.