Machine-Learning Microfluidic Minute-Scale Microorganism Metrics Monitoring(M6).

Area of Science:

• Microbiology
• Biosensing Technology
• Computational Biology

Background:

• On-site monitoring of airborne microorganisms, especially pathogens, is hindered by low concentrations and background interference.
• Existing methods struggle with the dynamic nature of aerosol diffusion, limiting real-time detection capabilities.

Purpose of the Study:

• To develop a rapid, on-site detection platform for airborne microorganisms.
• To integrate microfluidic separation, electrochemical impedance spectroscopy, and machine learning for enhanced analysis.
• To demonstrate the platform's efficacy using African swine fever virus (ASFV) as a model.

Main Methods:

• A Puri-focusing microfluidic chip designed for fluid-dynamic focusing achieved high separation efficiency for target particles.
• Electrochemical impedance spectroscopy (EIS) was employed to extract impedance features from aerosol samples.
• Machine learning classifiers, including Random Forest (RF), were utilized for data analysis and pathogen classification.

Main Results:

• The microfluidic chip demonstrated 95.8% separation efficiency for 5 µm particles.
• The Random Forest model achieved 95.2% classification accuracy for airborne microorganism analysis.
• The platform achieved a detection limit of 188 TCID50/mL for ASFV in air samples, with rapid detection cycles (<1 minute).

Conclusions:

• The integrated platform offers a feasible solution for rapid on-site monitoring of aerosol-transmitted microorganisms.
• This technology has significant potential applications in public health, agriculture, and production safety.
• The combination of microfluidics, EIS, and machine learning provides a powerful approach for biosensing challenges.