AI-Assisted Multiplex SYBR Green I-Based qPCR for the Identification and Quantitative Analysis of Oral Microbiota

Conventional multiplex qPCR faces challenges in quantitative multipathogen detection due to limited fluorescent channels, primer interference, and the high cost of probes. This study developed a novel velocity-controlled PCR (VC-PCR) system integrated with artificial intelligence (AI) for high-throughput, quantitative pathogen detection. The system was optimized with bovine serum albumin (BSA) and high-concentration SYBR Green I, enabling the use of standard primers without probes. Key parameters like BSA, dye, primer concentrations, and product lengths were systematically tested. An AI framework, employing a combination of algorithms including XGBoost, numerical interpolation, and linear regression, was used for duplex assays. Another AI framework based on the varied k-NN (k = 1) algorithm combining ShapeDTW and RMSE was designed for triplex assays to interpret complex amplification curves. Optimal conditions were determined as 5 mg/mL BSA, 8× SYBR Green I, and 500 nM total primers. This setup clearly differentiated Streptococcus mutans, Lactobacillus acidophilus, and Actinomyces viscosus, based on End RFU, slopes, and inflection points. With AI, duplex VC-PCR achieved 100% semiquantitative accuracy and a 3.7% mean absolute quantification error. Triplex VC-PCR attained 100% qualitative analysis and over 90% accuracy in semiquantitative analysis. The limits of detection (LOD) in triplex assays were 1.13-fold for A. viscosus, 1.20-fold for L. acidophilus, and 2.10-fold for S. mutans. The multiplex VC-PCR system eliminates the need for melt curve analysis or multiple fluorescence channels for real-time and high-throughput amplification monitoring and quantitative detection. It provides a cost-effective, high-sensitivity solution for the simultaneous detection of multiple pathogens with broad implications for clinical diagnostics, environmental monitoring, and public health.