Digital by design approach to develop a universal deep learning AI architecture for automatic chromatographic peak integration

Abstract

Chromatographic data processing has garnered attention due to multiple Food and Drug Administration 483 citations and warning letters, highlighting the need for a robust technological solution. The healthcare industry has the potential to greatly benefit from the adoption of digital technologies, but the process of implementing these technologies can be slow and complex. This article presents a "Digital by Design" managerial approach, adapted from pharmaceutical quality by design principles, for designing and implementing an artificial intelligence (AI)-based solution for chromatography peak integration process in the healthcare industry.We report the use of a convolutional neural network model to predict analytical variability for integrating chromatography peaks and propose a potential GxP framework for using AI in the healthcare industry that includes elements on data management, model management, and human-in-the-loop processes. The component on analytical variability prediction has a great potential to enable Industry 4.0 objectives on real-time release testing, automated quality control, and continuous manufacturing.

Automation in Analytical Chemistry: The Role of Ai in Chromatography

“…The authors, from Merck Serono (Italy) and Bosch Global Software Technologies Private Limited (India), proposed a potential GxP framework for using AI across the healthcare industry. The project was executed under a Digital Innovation Management framework, ensuring the involvement of stakeholders and decision-makers from proof of concept through proof of feasibility and finally to proof of value [41].…”

Automatic peak annotation and area estimation of glycan map peaks directly from chromatograms

“…Chromatography data systems (e.g., Chromeleon and Empower) usually offer an automated integration feature consisting of four major processes. 7 First, the automated integration feature identifies the peak by the apex and then moves downward and outward to determine the baseline around the peak. It proceeds to skimming each peak within a cluster with Gaussian profiles to determine its boundaries and afterward determines the peak's area, height, and retention time.…”

“…10 While several chemometric 11 and neural network 12 approaches have already been proposed, we have made use of a different, simpler, yet effective way of processing chromatograms centered around time warping and regression. Unlike convolutional neural networks 7 and other complex bottom-up solutions, 13 our targeted and supervised approach delivers more interpretable results and provides clear insight into the strengths of relationships between variables and overall statistical inference. Moreover, the proposed approach is more intuitive and simpler to communicate compared with black box solutions.…”