CIME4R: Exploring iterative, AI-guided chemical reaction optimization campaigns in their parameter space

CIME4R Teaser

Abstract

Chemical reaction optimization (RO) is an iterative process that results in large and high-dimensional datasets. Current tools only allow for limited analysis and understanding of parameter spaces, making it hard for scientists to review or follow changes throughout the process. With the recent emergence of using artificial intelligence (AI) models to aid RO, another level of complexity was added. It is critical to assess the quality of a model’s prediction and understand its decision to aid human-AI collaboration and trust calibration. To that regard, we propose CIME4R—an open-source interactive web application for analyzing RO data and AI predictions. CIME4R supports users in (i) comprehending a reaction parameter space, (ii) investigating how the RO process developed over iterations, (iii) identifying critical factors of a reaction, and (iv) understanding model predictions. This aids users in making informed decisions during the RO process and helps them review an RO process in retrospect, especially in the realm of AI-guided RO. CIME4R aids decision-making through the interaction between humans and AI by combining the strengths of expert experience and high computational precision. We developed and tested CIME4R together with domain experts and verified its usefulness with three case studies. With CIME4R the experts were able to produce valuable insights from past RO campaigns and make informed decisions on which experiments to perform next. We believe that CIME4R is the beginning of an open-source community project that improves the workflow of scientists working in the reaction optimization domain.


Citation

Christina Humer, Rachel Nicholls, Henry Heberle, Moritz Heckmann
Michael Pühringer, Thomas Wolf, Maximilian Lübbesmeyer, Julian Heinrich, Julius Hillenbrand, Giulio Volpin, Marc Streit
CIME4R: Exploring iterative, AI-guided chemical reaction optimization campaigns in their parameter space
ChemRxiv Preprint, doi:10.26434/chemrxiv-2023-218lq, 2023.

BibTeX

@article{2024_cime4r,
    title = {CIME4R: Exploring iterative, AI-guided chemical reaction optimization campaigns in their parameter space},
    author = {Christina Humer and Rachel Nicholls and Henry Heberle and Moritz Heckmann and Michael Pühringer and Thomas Wolf and Maximilian Lübbesmeyer and Julian Heinrich and Julius Hillenbrand and Giulio Volpin and Marc Streit},
    journal = {ChemRxiv Preprint},
    doi = {10.26434/chemrxiv-2023-218lq},
    url = {https://doi.org/10.26434/chemrxiv-2023-218lq},
    year = {2023}
}

Acknowledgements

This work was supported by the JKU Visual Data Science Lab and Bayer AG. We thank Niklas Hölter for supporting with early testing of CIME4R. We thank the participants of the case studies. We thank Dominic Girardi for UI testing.

Funding

This work was supported in part by Bayer AG, State of Upper Austria and the Austrian Federal Ministry of Education, Science and Research via the LIT – Linz Institute of Technology (LIT-2019-7-SEE-117), and the Austrian Science Fund (FWF DFH 23–N). Funding by the Life Science Collaboration Program of Bayer AG (Projects ”LSC MIC DROP” and ”Explainable AI” ) is gratefully acknowledged.