Enhancing Classification Robustness in Real-Time Digital PCR via Amplification Curve Regulation and Clustering Analysis

Real-time digital PCR technology holds significant promise in nucleic acid detection. However, amplification heterogeneity will hinder the effective separation of positive and negative amplification curves, leading to partition misclassification and a reduced quantitative precision. Herein, a fluorescence-based dynamic regulation (FIRM) and intrinsic-feature clustering method was proposed to regulate nucleic acid amplification in microporous chips and improve the analytical ability of amplification curves, aiming to enhance classification robustness in real-time digital PCR. By optimizing kinetic conditions and using the temperature-dependent dye 5-TAMRA for closed-loop temperature control within microwells, the FIRM method achieved more efficient amplification. Additionally, based on the maximum slope and coefficient of variation of each amplification curve, an intrinsic-feature clustering method was used to mitigate classification uncertainty. Compared to traditional amplification and analysis methods, the FIRM and intrinsic-feature clustering method improved the gap ratio at different concentrations, increasing from 65.3%, 7.8%, 3.5%, and 0.4% to 68.6%, 41.3%, 47.6%, and 22.9%, respectively, demonstrating enhanced classification robustness. This improvement enabled more effective differentiation between positive and negative amplification curves, thereby facilitating precise quantification in real-time digital PCR. The detection limit was 1 copy/test, and it was effectively applied to clinical sample analysis. This method holds great potential promise for advancing real-time digital PCR technology with potential applications in pathogen detection, disease diagnostics, and other biomedical fields.