Single-cell impedance flow cytometry (IFC) is emerging as a label-free and non-invasive method for characterizing the electrical properties and revealing sample heterogeneity. At present, most IFC studies utilize phenomenological parameters (e.g., impedance amplitude, phase and opacity) to characterize single cells instead of intrinsic biophysical metrics (e.g., radius r, cytoplasm conductivity σ_i and specific membrane capacitance C_sm). Intrinsic parameters are normally calculated off-line by time-consuming model-fitting methods. Here, we propose to employ neural network (NN)-enhanced IFC to achieve both real-time single-cell intrinsic characterization and intrinsic parameter-based cell classification at high throughput. Three intrinsic parameters (r, σ_i and C_sm) can be obtained online and in real-time via a trained NN at 0.3 ms per single-cell event, achieving significant improvement in calculation speed. Experiments involving four cancer cells and one lymphocyte cell demonstrated 91.5% classification accuracy in the cell type for a test group of 9751 cell samples. By performing a viability assay, we provide evidence that the IFC test per se would not substantially affect the cell property. We envision that the NN-enhanced real-time IFC will provide a new platform for high-throughput, real-time and online cell intrinsic electrical characterization.

Tumor-derived circulating cell-free DNA (cfDNA) provides critical clues for cancer early diagnosis, yet it often suffers from low sensitivity. Here, we present a cancer early diagnosis approach using tumor fractions deciphered from circulating cfDNA methylation signatures. We show that the estimated fractions of tumor-derived cfDNA from cancer patients increase significantly as cancer progresses in two independent datasets. Employing the predicted tumor fractions, we establish a Bayesian diagnostic model in which training samples are only derived from late-stage patients and healthy individuals. When validated on early-stage patients and healthy individuals, this model exhibits a sensitivity of 86.1% for cancer early detection and an average accuracy of 76.9% for tumor localization at a specificity of 94.7%. By highlighting the potential of tumor fractions on cancer early diagnosis, our approach can be further applied to cancer screening and tumor progression monitoring.