Machine Learning-Based Detection of Anaplasma spp. Using Dielectric Properties of Host Cells

Area of Science:

• Biophysics
• Cell Biology
• Infectious Diseases

Background:

• Tick-borne bacterial infections, like Anaplasma phagocytophilum, present diagnostic challenges due to reliance on specialized lab tests.
• Dielectric measurements offer a label-free method to assess cellular states, potentially aiding in early detection.

Purpose of the Study:

• To investigate the potential of dielectric measurements for distinguishing Anaplasma phagocytophilum-infected from uninfected HL-60 human promyelocytic leukemia cells.
• To evaluate the efficacy of specific dielectric properties (cytoplasmic conductivity, membrane conductance, membrane capacitance) in differentiating infection states.

Main Methods:

• In vitro dielectric measurements were performed on infected and uninfected HL-60 cells under controlled medium conductivities (100 and 300 µS/cm).
• Exploratory analysis, statistical tests, and linear classifiers were employed, with performance assessed via leave-one-out cross-validation.
• Key dielectric properties analyzed included cytoplasmic conductivity, specific membrane conductance, and specific membrane capacitance.

Main Results:

• Specific membrane conductance emerged as the most informative property for distinguishing between infected and uninfected cells.
• Higher medium conductivity (300 µS/cm) enhanced both classification performance and effect sizes.
• Membrane-related properties consistently provided the clearest separation between infection states across different models.

Conclusions:

• Dielectric measurements, particularly specific membrane conductance, show potential as a label-free method for detecting Anaplasma phagocytophilum infection in human cells.
• Preliminary findings suggest this approach could complement existing diagnostic methods for tick-borne diseases.
• Further research with larger sample sizes and diverse experimental conditions is warranted to validate these pilot results.