Current trends in chromatographic prediction using artificial intelligence and machine learning

Area of Science:

• Analytical Chemistry
• Computational Chemistry

Background:

• Artificial intelligence (AI) and machine learning (ML) are increasingly utilized for their predictive capabilities, accuracy, and speed across various scientific domains.
• In chromatography, AI and ML are particularly valuable for method development, offering efficient and accurate solutions for predicting chromatographic characteristics.

Purpose of the Study:

• To review various AI and ML models used for determining chromatographic characteristics.
• To explore artificial neural network (ANN) techniques and their advantages over classical linear models in liquid chromatography.
• To highlight the benefits of integrating fuzzy systems with ANNs and combining AI/ML with Quantitative Structure-Retention Relationships (QSRR) for enhanced prediction.

Main Methods:

• Review of existing literature on AI and ML applications in chromatography.
• Analysis of artificial neural network (ANN) based techniques for chromatographic prediction.
• Investigation of hybrid approaches, including fuzzy systems with ANNs and QSRR combined with ANNs.

Main Results:

• ANN-associated techniques demonstrate higher accuracy and potential for predicting chromatographic characteristics compared to classical linear models.
• Integration of fuzzy systems with ANNs provides more efficient and accurate chromatographic prediction methods.
• Combining AI/ML algorithms with QSRR significantly improves the accuracy of target molecule retention prediction.

Conclusions:

• AI and ML algorithms, especially ANNs and hybrid models, offer powerful tools for advancing chromatographic method development and prediction.
• These advanced computational approaches show significant potential for overcoming challenges in analytical chemistry, leading to more precise and efficient analyses.
• The integration of AI/ML with QSRR represents a promising direction for accurate retention prediction in liquid chromatography.