Establishing performance metrics for quantitative non-targeted analysis: a demonstration using per- and...

Area of Science:

• Environmental Chemistry
• Analytical Chemistry

Background:

• Non-targeted analysis (NTA) is increasingly used for identifying unknown chemicals.
• Quantitative NTA (qNTA) relies on calibration surrogates, potentially reducing accuracy compared to targeted methods.

Purpose of the Study:

• To evaluate and compare the performance of targeted analysis and qNTA.
• To introduce new metrics for assessing predictive accuracy, uncertainty, and reliability in quantitative chemical analysis.

Main Methods:

• Developed and applied novel metrics for accuracy, uncertainty (95% inverse confidence intervals), and reliability.
• Compared five quantitative approaches, including targeted analysis and qNTA with "global" and "expert-selected" surrogates.
• Analyzed a mixture of 29 per- and polyfluoroalkyl substances (PFAS).

Main Results:

• Targeted approaches demonstrated superior performance, especially with matched calibration and internal standards.
• The most generalizable qNTA approach showed ~4x lower accuracy, ~1000x higher uncertainty, and ~5% lower reliability than targeted analysis.
• Using "expert-selected" surrogates improved qNTA accuracy and uncertainty but decreased reliability.

Conclusions:

• qNTA is useful for analyzing certain emerging contaminants, but performance lags behind targeted methods.
• The developed metrics provide a framework for evaluating and improving quantitative analytical methods.
• Further research is needed to enhance qNTA for more complex chemical mixtures.