Machine learning-assisted nanosensor arrays: An efficiently high-throughput food detection analysis

Background: How to timely identify the food quality through a low-cost, easy operation, and high-throughput way is a milestone protects for food industry, especially in resource-limited area. Nanosensors by integrating with biomolecules (such as antibodies and aptamers) have emerged substitutes for the standard equipment analysis in the large-scale screening. However, the expensive cost of biomolecules, the "lock-key" combination unable to solve some problems (such as food freshness), and emerging food risks have strictly their development in food industry. Additionally, nanosensors without biomolecules are easily suffered from the non-specific interference, making the detection results unreliable. Therefore, some studies have concentrated on the sensor array by using nanomaterials as receptors to solve abovementioned problems, which is based on the multiple signal responses to generate the distinctive fingerprint for each analyte. Scope and approach: This review comprehensively discussed the machine learning-assisted nanosensor arrays for the efficiently high-throughput food detection analysis, which mainly concludes candidates for nanosensor arrays, commonly used machine learning algorithms, and the application in food applications (such as foodborne hazards, food components, food freshness, food origin, and food adulteration). Additionally, we have proposed the challenges and prospects of machine learning-assisted nanosensor arrays in food applications to bridge the gap of current development bottleneck. Key findings and conclusions: Therefore, machine learning-assisted nanosensor arrays for the efficiently highthroughput detection analysis in food industry are proposed.