P-10 Machine Learning Assisted ECL Sensor Analysis
Presenter Status
Chair and Professor of Dept. of Engineering
Second Presenter Status
Associate professor, Dept. of Computing
Third Presenter Status
Visiting scholar, Dept. of Andrews University
Fourth Presenter Status
Student, Dept. of Computing
Fifth Presenter Status
Student, Dept. of theology
Sixth Presenter Status
Professor, Dept. of Publich health
Preferred Session
Poster Session
Start Date
25-10-2019 2:00 PM
Presentation Abstract
Current biosensor data analysis is based on calibration curves, which are generated based on a single key feature of the sensor signal, while other features in the biosensor are generally ignored. Inclusion of these ignored features allows us to make more informed predictions regarding unknown concentrations. In this work, a multimodal electrochemiluminescent (ECL) sensor was explored to improve on the data analysis using machine learning (ML) techniques. The multimodal ECL sensor data was collected for light intensity and electrochemical data in the Ru(bpy)2(3+)/tri-n-propylamine system, and a disposable electrode was used as a sensor unit. The results from ML based analysis demonstrate the potential to replace the traditional calibration curve based method and have potential to improve development of low-cost point-of-care diagnostic devices.
P-10 Machine Learning Assisted ECL Sensor Analysis
Current biosensor data analysis is based on calibration curves, which are generated based on a single key feature of the sensor signal, while other features in the biosensor are generally ignored. Inclusion of these ignored features allows us to make more informed predictions regarding unknown concentrations. In this work, a multimodal electrochemiluminescent (ECL) sensor was explored to improve on the data analysis using machine learning (ML) techniques. The multimodal ECL sensor data was collected for light intensity and electrochemical data in the Ru(bpy)2(3+)/tri-n-propylamine system, and a disposable electrode was used as a sensor unit. The results from ML based analysis demonstrate the potential to replace the traditional calibration curve based method and have potential to improve development of low-cost point-of-care diagnostic devices.
Acknowledgments
This research was funded by the National Science Foundation (NSF) (Grant number 1706597).