Quantification of 14 Major and Minor Cannabinoids with Absorbance-Transmittance Excitation-Emission Matrix Spectroscopy and Machine Learning.

BACKGROUND: Cannabis plants (Cannabis sativa) contain a diverse group of terpenophenolic compounds known as phytocannabinoids, with 131 cannabinoids identified to date. Rapid and low-cost analytical approaches capable of quantifying both major and minor cannabinoids are increasingly important for research, quality control, and regulatory applications. This study evaluates the patented Absorbance-Transmittance Excitation-Emission Matrix (A-TEEM™) spectroscopic technique as a fast and reliable alternative to conventional chromatographic methods. A-TEEM integrates ultraviolet-visible absorbance and fluorescence measurements while correcting for absorbance-dependent inner-filter effects, enabling linear relationships between fluorescence intensity and analyte concentration. The primary objective was to calibrate and validate machine learning models using A-TEEM data for cannabinoid quantification, benchmarked against a validated high-performance liquid chromatography-photodiode array (HPLC-PDA) reference method. MATERIALS AND METHODS: A total of 49 dry cannabis flower extracts were analyzed using the A-TEEM technique to quantify 14 cannabinoids. Spectral data generated by A-TEEM were directly compared with concentration data obtained from an established and validated HPLC-PDA method. Extreme gradient boosting regression models were developed using HPLC-PDA results as reference values to predict cannabinoid concentrations from A-TEEM spectral data and to evaluate quantitative performance. RESULTS: The A-TEEM method demonstrated rapid, robust, and sensitive quantification of all 14 target cannabinoids. Model performance metrics, including coefficients of determination (R2) and limits of detection (LOD) and limits of quantification (LOQ), are scaled proportionally with the maximum cannabinoid concentrations present in the samples. For major cannabinoids exceeding 0.35% concentration, the mean combined cross-validation and validation R2 reached 0.994 ± 0.005, with mean LOD and LOQ values of 0.0146% and 0.0442%, respectively. Cannabinoids present between 0.35% and 0.1% showed mean LOD/LOQ values of 0.00278% and 0.00842%, while minor cannabinoids below 0.1% exhibited even lower LOD/LOQ values of 0.0004% and 0.00128%, respectively. In addition, A-TEEM concentration profiles enabled clear qualitative and quantitative differentiation of three cannabis chemovars: tetrahydrocannabinol (THC)-dominant, cannabidiol (CBD)-dominant, and THC-CBD-intermediate hybrids. CONCLUSIONS: The A-TEEM technique provides a sensitive, rapid, and cost-effective approach for the qualitative and quantitative determination of both major and minor cannabinoids in solution. Its analytical performance is comparable to that of the reference HPLC-PDA method while offering substantial advantages in speed, simplicity, and suitability for high-throughput analysis.