Compensation for Matrix Effects in High-Dimensional Data using Standard Addition

Itai Dattner, University of Haifa, Haifa, Israel (idattner@stat.haifa.ac.il)
Israel Schechter, Technion, Haifa, Israel
Elena Khanonkin, Technion, Haifa, Israel

A common method for compensation for matrix effects in analytical chemistry is application of the standard addition method. It works well when utilizing a single signal (e.g., absorbance at a single wavelength). However, high-dimensional data are often available (e.g, the spectrum in a given range), which could be utilized for improving the results. We propose an algorithm that compensates for matrix effects, when high-dimensional data are available, such that all data is involved in the standard addition, allowing for better results. The algorithm is based on construction of a PCR model for the analyte without the matrix, performing standard addition at all signals (e.g., wavelengths) and thus achieving a corrected set of data (e.g., spectra), and finally, applying the PCR model to the corrected data. The algorithm was tested, and its performance was evaluated. The effects of the SNR and the degree of the matrix effects were investigated. It was found that the algorithm works better than direct PCR and compensates well for matrix effects.  

---

Short Biography of Presenting Author

Dr. Itai Dattner is the founder and head of the Scientific Machine Learning (SciML) Research Lab at the University of Haifa’s Department of Statistics. His research bridges the realms of machine learning, artificial intelligence, dynamical systems, and statistical methodologies to tackle complex, real-world challenges. The lab engages in both basic and applied research, with a strong emphasis on interdisciplinary collaboration across fields such as medicine, epidemiology, psychotherapy, agriculture, and engineering.