Large language models (LLMs) have demonstrated impressive capabilities, but the bar for clinical applications is high. Attempts to assess the clinical knowledge of models typically rely on automated evaluations based on limited benchmarks. Here, to address these limitations, we present MultiMedQA, a benchmark combining six existing medical question answering datasets spanning professional medicine, research and consumer queries and a new dataset of medical questions searched online, HealthSearchQA. We propose a human evaluation framework for model answers along multiple axes including factuality, comprehension, reasoning, possible harm and bias. In addition, we evaluate Pathways Language Model1 (PaLM, a 540-billion parameter LLM) and its instruction-tuned variant, Flan-PaLM2 on MultiMedQA. Using a combination of prompting strategies, Flan-PaLM achieves state-of-the-art accuracy on every MultiMedQA multiple-choice dataset (MedQA3, MedMCQA4, PubMedQA5 and Measuring Massive Multitask Language Understanding (MMLU) clinical topics6), including 67.6% accuracy on MedQA (US Medical Licensing Exam-style questions), surpassing the prior state of the art by more than 17%. However, human evaluation reveals key gaps. To resolve this, we introduce instruction prompt tuning, a parameter-efficient approach for aligning LLMs to new domains using a few exemplars. The resulting model, Med-PaLM, performs encouragingly, but remains inferior to clinicians. We show that comprehension, knowledge recall and reasoning improve with model scale and instruction prompt tuning, suggesting the potential utility of LLMs in medicine. Our human evaluations reveal limitations of today’s models, reinforcing the importance of both evaluation frameworks and method development in creating safe, helpful LLMs for clinical applications.

Chiral ferrocene ligands have been widely used in asymmetric catalysis. The advantages of using ferrocene as a scaffold for chiral ligands are described, particularly those regarding planar chirality, rigid bulkiness, and ease of derivatization. The role of planar chirality in 1,2- and 1,1'-disubstituted ferrocene systems is discussed. By using a bulky ferrocene fragment, novel ferrocene ligands were designed, and high enantioselectivity and regioselectivity were achieved in the allylic substitution reaction of monosubstituted allyl substrates. Using the tunable electronic properties of a diphosphine-oxazoline ferrocenyl ligand, the regioselectivity of the intermolecular asymmetric Heck reaction was also examined.

No abstract

Over the past two decades, supramolecular gels have attracted significant attention from scientists in diverse research fields and have been extensively developed. This review mainly focuses on the significant achievements in supramolecular gels and catalysis. First, by incorporating diverse catalytic sites and active organic functional groups into gelator molecules, supramolecular gels have been considered as a novel matrix for catalysis. In addition, these rationally designed supramolecular gels also provide a variety of templates to access metal nanocomposites, which may function as catalysts and exhibit high activity in diverse catalytic transformations. Finally, as a new kind of biomaterial, supramolecular gels formed in situ by self-assembly triggered by catalytic transformations are also covered herein.