.png){kind=icon}
Return
OFAT-BO: A one-factor-at-a-time Bayesian optimization method for continuous-flow experiment optimization
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ces.2026.124607.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Bayesian optimization is attractive for reaction optimization, but many laboratory implementations assume rapid feedback between experiments and model updates. In continuous-flow studies that rely on offline characterization, one-point-per-decision-cycle Bayesian optimization can become operationally inefficient because each analytical cycle incurs fixed setup and completion overhead. Herein, we present OFAT-BO, a workflow-constrained batch Bayesian optimization strategy that selects five candidate conditions per decision cycle while varying only one factor within each batch. This design is intended to reduce characterization cycles and condition-switching overhead under offline analytics while preserving Bayesian model updates. Benchmark studies and photochemical flow case studies show that OFAT-BO can achieve higher empirical success rates than a sequential Bayesian optimization baseline. In a photocatalytic C-N cross-coupling case study, OFAT-BO identified a high-yielding condition (91.34 %) and revealed trends consistent with coupled catalyst-concentration and light-attenuation effects. These results support OFAT-BO as a practical optimization strategy for continuous-flow experiments performed with offline characterization.
Keywords:
Bayesian optimization
Continuous-flow experiment
Offline characterization
Photochemistry
C -N cross-coupling
Journal
IF:
4.3
Papers:
2.3W
Citations:
5.5W
Organization
Cited Papers
No cited papers available