Clustered Regularly Interspaced Short Palindromic Repeat-Based Colorimetric Aptasensor Combined with Smartphone Imaging and Deep Learning Enables Selective Recycling and Visual Prediction of Microplastics in the Environment

Microplastics present significant risks to human health and ecosystem stability, creating an urgent need for analytical methods that are simple, rapid, sensitive, and field-deployable. Herein, we report a clustered regularly interspaced short palindromic repeat (CRISPR)-based colorimetric aptasensor for the detection of poly(vinyl chloride) (PVC) and polystyrene (PS) microplastics. This platform leverages the high specificity of PVC and PS aptamers integrated into a Fe 3 O 4 @Au-DNA magnetic complex, which facilitates capture, separation, and detection. Upon microplastic binding, a competitive reaction releases an activator DNA, initiating a dual CRISPR-Cas12a system for signal amplification. The activated Cas12a trans-cleavage activity is then linked to a hemin-aptamer DNAzyme colorimetric reaction, converting the signal into a visible color change. This colorimetric output is captured by smartphone imaging and processed in real time. Furthermore, a deep-learning-based regression model was developed to enable the quantitative prediction of PVC and PS micro/nanoplastics in diverse environmental matrices. The method exhibited high selectivity and a broad dynamic range from 10 –2 to 10 3 μg/mL. In smartphone detection mode, the limits of detection for PVC and PS reached 3.1 ng/mL and 3.7 ng/mL, respectively. This approach significantly enhances detection performance and stability, enabling visual monitoring of microplastics in complex real samples. Collectively, this work provides a rapid and effective strategy for the extraction and real-time quantification of small molecules.