Reduction of Biosensor False Responses and Time Delay Using Dynamic Response and Theory-Guided Machine Learning.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Biosensing Technology

Background:

• Biosensors face challenges with false results and time delays, limiting their widespread adoption in industrial, healthcare, military, and consumer sectors.
• Traditional methods like standard curve regression often fall short in optimizing biosensor accuracy and speed.

Purpose of the Study:

• To develop and validate a novel methodology for reducing false results and time delay in biosensors.
• To improve biosensor accuracy and speed by integrating machine learning with domain-specific knowledge.

Main Methods:

• Utilized machine learning integrated with biosensing domain knowledge for feature engineering.
• Employed theory-guided feature engineering and compared it with traditional methods (TSFRESH).
• Validated the methodology using cantilever biosensors for microRNA quantification via dynamic response analysis.

Main Results:

• The new methodology significantly improved biosensor accuracy and speed compared to traditional regression analysis.
• Theory-guided feature engineering enhanced classification model performance and efficiency.
• Rapid and accurate quantification of microRNA was achieved using the initial transient response, reducing data acquisition time.

Conclusions:

• Integrating explainable machine learning with domain knowledge offers a powerful approach to overcome key biosensor limitations.
• The methodology enables faster, more accurate biosensing and can guide experimental design for optimal performance.
• This work paves the way for improved biosensor applications by reducing false results and time delays.