Commit: Automated Scale-Up Crystallisation DataFactory for model-based pharmaceutical process development: a data-driven, mechanistic and hybrid model case study
Abstract
In 2025, we introduced the Scale-Up Crystallisation DataFactory, an automated platform for crystallisation experimentation and Bayesian optimisation. However, the initial case study used a relatively simple three-parameter, single-objective optimisation problem. In this Commit paper, we extend the platform to a five-dimensional, multi-objective cooling-antisolvent crystallisation of lovastatin and benchmark data-driven, mechanistic and hybrid optimisation approaches. Sixteen screening experiments were followed by model-guided experiments targeting yield and particle size (D90). Bayesian optimisation produced continued improvements in hypervolume, while mechanistic and hybrid approaches did not improve experimental performance beyond screening. The results demonstrate the value of experimentally benchmarking optimisation strategies for autonomous process development.
Related
- Automated transition state generation for mechanistic exploration in organic synthesis
- Development and validation of a GC-MS method for the determination of ethylene glycol and diethylene glycol in pharmaceutical syrups.
- Direct Access to Carbenoids from Alkyl and α-Acyloxy Halide`s for Boron Homologations: Method Development and Mechanistic Insights
- Ti3C2Tx MXene-Supported Anisotropic Crystallisation of Solvent-Casted p(VDF-TrFE) Films for Improved Elastic Deformation and Piezoelectric Response
- Substrate-induced crystallisation in polymers: Exceptional impact of prefreezing versus heterogeneous nucleation on kinetics and morphology
BibTeX
@misc{pickles2026commit,
title = {{Commit: Automated Scale-Up Crystallisation DataFactory for model-based pharmaceutical process development: a data-driven, mechanistic and hybrid model case study}},
author = {Thomas Pickles and Cameron J. Brown},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010269/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010269/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Commit: Automated Scale-Up Crystallisation DataFactory for model-based pharmaceutical process development: a data-driven, mechanistic and hybrid model case study
AU - Thomas Pickles
AU - Cameron J. Brown
PY - 2026
DA - 2026/10/12
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010269/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010269/v1
AB - In 2025, we introduced the Scale-Up Crystallisation DataFactory, an automated platform for crystallisation experimentation and Bayesian optimisation. However, the initial case study used a relatively simple three-parameter, single-objective optimisation problem. In this Commit paper, we extend the platform to a five-dimensional, multi-objective cooling-antisolvent crystallisation of lovastatin and benchmark data-driven, mechanistic and hybrid optimisation approaches. Sixteen screening experiments were followed by model-guided experiments targeting yield and particle size (D90). Bayesian optimisation produced continued improvements in hypervolume, while mechanistic and hybrid approaches did not improve experimental performance beyond screening. The results demonstrate the value of experimentally benchmarking optimisation strategies for autonomous process development.
ER -