Faculty Mentor Information
Dr. John Kalivas (Mentor), Idaho State University
Additional Funding Sources
This material was based upon work supported by the National Science Foundation under Grant No. 2305020.
Abstract
Within analytical chemistry, it is increasingly difficult to determine the best method to predict analyte values of target samples from expansive datasets. These datasets contain many different samples, each with unique underlying matrix effects. For a target sample to be predicted accurately, the source samples used to build a model must be both matrix effect (spectrally) matched and contain similar analyte values. To achieve this goal, an algorithm named local adaptive fusion regression (LAFR) was recently developed. This algorithm, while already effective, was modified to utilize a method of determining similarity called physicochemical responsive integrated similarity measure (PRISM). Using PRISM z-scoring throughout the LAFR algorithm allows for prediction reliability to be assessed, while maintaining the efficacy of the original algorithm. Additionally, these changes pave the way for future use in immersive virtual reality (IVR). Implementing IVR will allow the user to leverage ingrained pattern recognition skills to make complex judgements that the LAFR algorithm cannot formulate due to its autonomous structure. Thus, using the principles of LAFR in IVR, the performance and versatility of LAFR as a whole will increase.
Local Multivariate Calibration Method Utilizing PRISM with Future Application in Virtual Reality
Within analytical chemistry, it is increasingly difficult to determine the best method to predict analyte values of target samples from expansive datasets. These datasets contain many different samples, each with unique underlying matrix effects. For a target sample to be predicted accurately, the source samples used to build a model must be both matrix effect (spectrally) matched and contain similar analyte values. To achieve this goal, an algorithm named local adaptive fusion regression (LAFR) was recently developed. This algorithm, while already effective, was modified to utilize a method of determining similarity called physicochemical responsive integrated similarity measure (PRISM). Using PRISM z-scoring throughout the LAFR algorithm allows for prediction reliability to be assessed, while maintaining the efficacy of the original algorithm. Additionally, these changes pave the way for future use in immersive virtual reality (IVR). Implementing IVR will allow the user to leverage ingrained pattern recognition skills to make complex judgements that the LAFR algorithm cannot formulate due to its autonomous structure. Thus, using the principles of LAFR in IVR, the performance and versatility of LAFR as a whole will increase.