Rapid Detection of Authenticity and Quality of Chinese Medicinal Materials Based on NIR Organic Sensor and Artificial Intelligent Algorithms

The quality of Chinese medicinal materials (CMMs) on the market varies greatly, particularly with respect to their authenticity and quality. The treatment of CMMs has been strictly regulated by the government (Chinese Pharmacopeia), e.g., sulfite content of the specific sulfur-treated CMMs should not exceed 400 ppm. However, testing sulfur content in sulfur-treated CMMs and food products is currently obtained through the slow and costly process of sending specimens to a laboratory where they had to be destroyed for the measurements to be taken. At present, the testing can only be conducted for spot checks on random samples. Spectral analysis techniques offer a non-destructive and rapid analytical approach that provides the non-intrusive testing of the chemical composition of food, CMMs and other organic matter. The quantitative analysis is achieved by measuring the spectral fingerprints of the samples and establishing a calibration model with the spectral information of quality traits. However, the spectra measured for the CMMs can be influenced by different factors, e.g., noises associated with measurement conditions or ambient light, and excessive redundancy of variables in the spectra, which determine the performance of the prediction model. Therefore, it is important to extract valuable information, particularly for identification of spectral fingerprints in, e.g., CMMs, that are closely related to the target metrics for generating a more accurate prediction model.

This collaborative project proposes to carry out interdisciplinary research to develop a practical solution for accurately measuring spectral fingerprint absorption of CMMs through developing a low power handheld rapid detection solution. A mechanistic study of changes in CMMs quality traits including adulteration, origin, age, sulfur fumigation and molecular regulations during the process and postharvest treatments of the CMMs will be carried out in parallel with the identification of spectral fingerprints. We will study the correlation between the NIR absorption of the CMM quality traits, e.g., texture, chemical compositions, for application in non-invasive CMM quality detection and authentication. The resulting technology will enable the users to make accurate decisions with immediate, tangible, and traceable messages through an interconnected ecosystem of real-time data integrated with IoT solutions.