Carrier biomaterials used in single-cell analysis face a bottleneck in protein detection sensitivity, primarily attributed to elevated false positives caused by nonspecific protein adsorption. Toward carrier biomaterials with ultra-low nonspecific protein adsorption, a self-evolving discovery is developed to address the challenge of high-dimensional parameter spaces. Automation across nine self-developed or modified workstations is integrated to achieve a "can-do" capability, and develop a synergy-enhanced Bayesian optimization algorithm as the artificial intelligence brain to enable a "can-think" capability for small-data problems inherent to time-consuming biological experiments, thereby establishing a self-evolving discovery for carrier biomaterials. Through this approach, carrier biomaterials with an ultra-low nonspecific protein adsorption index of 0.2537 are successfully discovered, representing an over 80% decrease, while achieving a 10 000-fold reduction in experiment workload. Furthermore, the discovered biomaterials are fabricated into microfluidic-used carriers for protein-analysis applications, showing a 9-fold enhancement in detection sensitivity compared to conventional carriers. This is the very demonstration of a self-evolving discovery for carrier biomaterials, paving the way for advancements in single-cell protein analysis and further its integration with genomics and transcriptomics.
Keywords: artificial intelligence; automated experiments; biomaterials; single‐cell analysis.
© 2025 Wiley‐VCH GmbH.