This review summarizes microfluidic programmable strategies for channels and flow, referring to microvalves, micropumps, digital microfluidics, multiplexers, micromixers, slip-, and block-based configurations.

Terahertz spectroscopy (0.1~10 THz), as a new type of non-destructive testing method with both microwave and infrared characteristics, has shown remarkable potential in the field of food quality testing in recent years. Its unique penetration, high sensitivity, and low photon energy characteristics, combined with chemometrics and machine learning methods, provide an efficient solution for the qualitative and quantitative analysis of complex food ingredients. In this paper, we systematically review the principles of terahertz spectroscopy and its key applications in food testing, focusing on its research progress in pesticide residues, additives, biotoxins, and mold, adulteration identification, variety identification, and nutrient content detection. By integrating spectral data preprocessing, reconstruction algorithms, and machine learning model optimization strategies, this paper further analyzes the advantages and challenges of this technology in enhancing detection accuracy and efficiency. In addition, combined with the urgent demand for fast and nondestructive technology in the field of food detection, the future development direction of the deep integration of terahertz spectroscopy technology and artificial intelligence is envisioned, with a view to providing theoretical support and technical reference for food safety assurance and nutritional health research.

The degree of harm caused by pesticide residues is closely related to its type and concentration. Current enzyme biosensor technology used for pesticide residue detection has limitations, including lacking a built-in heating mechanism and an integrated platform for electrical impedance measurement technology to perform qualitative and quantitative analysis. This study proposes a high-throughput detection integrated platform and method based on enzyme sensors. Firstly, a double-layer electrode structure was developed to accommodate enzyme-inhibited biological reactions and temperature control functions. Subsequently, a high-throughput integrated detection platform was established, targeting three organophosphorus pesticides: dichlorvos, diazinon, and malathion. Time-series impedance data obtained during the enzyme inhibition reactions were used as features, combined with machine learning to analyze pesticide residues. The classification decision tree identified pesticide types with a 96.3% accuracy rate. The regression decision tree predicted the concentration range of pesticide residues (0.1mg/kg) with an average coefficient of determination of 0.953. Moreover, this method can analyze 32 samples within 15 minutes. Finally, accurate sample analyses were conducted and compared with gas chromatography, demonstrating good consistency. This approach meets the rapid detection requirements of the pesticide residue testing market and offers significant applicability and practicality.