The outbreak of emerging acute viral diseases urgently requires the acceleration of specialized antiviral drug development, thus widely adopting phenotypic screening as a strategy for drug repurposing in antiviral research. However, traditional phenotypic screening methods typically require several days of experimental cycles and lack visual confirmation of a drug's ability to inhibit viral infection. Here, we report a robust method that utilizes quantum-dot-based single-virus tracking and machine learning to generate unique single-virus infection fingerprint data from viral trajectories and detect the dynamic changes in viral movement following drug administration. Our findings demonstrated that this approach can successfully identify viral infection patterns at various infection phases and predict antiviral drug efficacy through machine learning within 90 min. This method provides valuable support for assessing the efficacy of antiviral drugs and offers promising applications for responding to future outbreaks of emerging viruses.
Read full abstract