Computing sensitivity and selectivity in parallel factor analysis and related multiway techniques: the need for...

Area of Science:

• Analytical Chemistry
• Chemometrics

Background:

• Sensitivity and selectivity are key metrics in multiway analysis for method comparison and experimental design.
• The second-order advantage in multiway analysis enables analyte determination despite complex backgrounds and interferences.

Purpose of the Study:

• To address the lack of general theory for multiway sensitivity estimation.
• To develop and assess Monte Carlo methods for evaluating sensitivity and selectivity in multiway techniques like PARAFAC.
• To extend net analyte signal theory for improved multianalyte and higher-order data analysis.

Main Methods:

• Utilized Monte Carlo numerical calculations to estimate variance inflation factors.
• Applied these methods to assess sensitivity and selectivity for parallel factor (PARAFAC) analysis and related multiway techniques.
• Investigated the limitations of existing net analyte signal theory for multianalyte and third-order data.

Main Results:

• Developed a robust approach using Monte Carlo simulations for estimating multiway sensitivity and selectivity.
• Demonstrated the utility of variance inflation factors for assessing analytical performance.
• Identified shortcomings in current net analyte signal theory for complex multiway scenarios.

Conclusions:

• The proposed Monte Carlo approach provides a practical tool for estimating multiway sensitivity and selectivity.
• Findings highlight the need for extending net analyte signal theory to accommodate multianalyte and higher-order data.
• The results have significant implications for optimizing the planning of multiway analytical experiments.