ABSTRACT
Custom-made, point-of-care PCR platforms are a necessary tool for rapid, point-of-care diagnostics in situations such as the current Covid-19 pandemic. However, a common issue faced by them is noisy fluorescence signals, which consist of a drifting baseline or noisy sigmoidal curve. This makes automated detection difficult and requires human verification. In this paper, we have tried to use nonlinear fitting for automated classification of PCR waveforms to identify whether amplification has taken place or not. We have presented several novel signal reconstruction techniques based on nonlinear fitting which will enable better pre-processing and automated differentiation of a valid or invalid PCR amplification curve. We have also tried to perform this classification at lower PCR cycles to reduce decision times in diagnostic tests.