Lecture
Digitally Compatible Classification and Automated Selection of HPLC Methods for Self-Driving Laboratories
- at -
- ICM Saal 3
- Type: Lecture
Lecture description
Pascal Miéville, Ngoc Van Thanh Nguyen, Jean-Charles Cousty, N. Bui, Guillaume Chanel, Cyril Portmann
High-throughput experimentaBon in chemistry (HTE) enables the parallel execuBon of hundreds of reacBons, but the rapid and robust selecBon of suitable high-performance liquid chromatography (HPLC) methods for product screening remains a major boLleneck. This
challenge was highlighted in an automated laboratory seOng conducBng over 300 reacBons in parallel, where subopBmal method selecBon led to substanBal Bme and material losses, ulBmately limiBng the overall efficiency and impact of HTE workflows. To address this issue, we developed an ensemble of machine-learning (ML)–based tools for the digital classificaBon of solid–liquid phase combinaBons. Leveraging readily accessible molecular descriptors, the framework automaBcally idenBfies and selects the most appropriate analyBcal condiBons, thereby improving decision-making speed, robustness, and resource efficiency in highthroughput HPLC screening.
challenge was highlighted in an automated laboratory seOng conducBng over 300 reacBons in parallel, where subopBmal method selecBon led to substanBal Bme and material losses, ulBmately limiBng the overall efficiency and impact of HTE workflows. To address this issue, we developed an ensemble of machine-learning (ML)–based tools for the digital classificaBon of solid–liquid phase combinaBons. Leveraging readily accessible molecular descriptors, the framework automaBcally idenBfies and selects the most appropriate analyBcal condiBons, thereby improving decision-making speed, robustness, and resource efficiency in highthroughput HPLC screening.
