The remarkable capabilities of Large Language Models (LLMs) make them increasingly compelling for adoption in real-world healthcare applications. However, the risks associated with using LLMs in medical applications have not been systematically characterized. We propose using five key principles for safe and trustworthy medical AI: Truthfulness, Resilience, Fairness, Robustness, and Privacy, along with ten specific aspects. Under this comprehensive framework, we introduce a novel MedGuard benchmark with 1,000 expert-verified questions. Our evaluation of 11 commonly used LLMs shows that the current language models, regardless of their safety alignment mechanisms, generally perform poorly on most of our benchmarks, particularly when compared to the high performance of human physicians. Despite recent reports indicate that advanced LLMs like ChatGPT can match or even exceed human performance in various medical tasks, this study underscores a significant safety gap, highlighting the crucial need for human oversight and the implementation of AI safety guardrails.

Information retrieval (IR) is essential in biomedical knowledge acquisition and clinical decision support. While recent progress has shown that language model encoders perform better semantic retrieval, training such models requires abundant query-article annotations that are difficult to obtain in biomedicine. As a result, most biomedical IR systems only conduct lexical matching. In response, we introduce MedCPT, a first-of-its-kind Contrastively Pre-trained Transformer model for zero-shot semantic IR in biomedicine. To train MedCPT, we collected an unprecedented scale of 255 million user click logs from PubMed. With such data, we use contrastive learning to train a pair of closely-integrated retriever and re-ranker. Experimental results show that MedCPT sets new state-of-the-art performance on six biomedical IR tasks, outperforming various baselines including much larger models such as GPT-3-sized cpt-text-XL. In addition, MedCPT also generates better biomedical article and sentence representations for semantic evaluations. As such, MedCPT can be readily applied to various real-world biomedical IR tasks.

Capturing sentence semantics plays a vital role in a range of text mining applications. Despite continuous efforts on the development of related datasets and models in the general domain, both datasets and models are limited in biomedical and clinical domains. The BioCreative/OHNLP organizers have made the first attempt to annotate 1,068 sentence pairs from clinical notes and have called for a community effort to tackle the Semantic Textual Similarity (BioCreative/OHNLP STS) challenge. We developed models using traditional machine learning and deep learning approaches. For the post challenge, we focus on two models: the Random Forest and the Encoder Network. We applied sentence embeddings pre-trained on PubMed abstracts and MIMIC-III clinical notes and updated the Random Forest and the Encoder Network accordingly. The official results demonstrated our best submission was the ensemble of eight models. It achieved a Person correlation coefficient of 0.8328, the highest performance among 13 submissions from 4 teams. For the post challenge, the performance of both Random Forest and the Encoder Network was improved; in particular, the correlation of the Encoder Network was improved by ~13%. During the challenge task, no end-to-end deep learning models had better performance than machine learning models that take manually-crafted features. In contrast, with the sentence embeddings pre-trained on biomedical corpora, the Encoder Network now achieves a correlation of ~0.84, which is higher than the original best model. The ensembled model taking the improved versions of the Random Forest and Encoder Network as inputs further increased performance to 0.8528. Deep learning models with sentence embeddings pre-trained on biomedical corpora achieve the highest performance on the test set.

PubMed queries frequently retrieve thousands of documents making it very challenging for a user to identify information of interest. In this paper we propose a method for automatically identifying central concepts in large PubMed retrievals. The centrality of concept is modeled using the hypergeometric distribution. Retrieved documents are grouped by concept, which can help users navigate the retrieval. We test our method on five datasets, each representing a medical condition.

We present a new algorithm, which we call PROBE, to find the minimum of a convex function. Such a minimization is important in many machine learning methods, including Support Vector Machines (SVM). We show that PROBE is a viable alternative to published algorithms for SVM learning with several important advantages. PROBE is a simple and easily programmed algorithm, with a well-defined, parametrized stopping criterion; it is not limited to SVM, but can be applied to other convex loss functions, such as the Huber and Maximum Entropy models; and its time and memory requirements are consistently modest in handling very large training sets.