LLM-Based Agents for Tool Learning: A Survey

Background and Aims Large language models (LLMs) have gained significant attention in the field of natural language processing (NLP), marking a shift from traditional techniques like Term Frequency-Inverse Document Frequency (TF-IDF). We developed a traditional NLP model to predict arteriovenous fistula (AVF) failure within next 30 days using clinical notes. The goal of this analysis was to investigate whether LLMs would outperform traditional NLP techniques, specifically in the context of predicting AVF failure within the next 30 days using clinical notes. Method We defined AVF failure as the change in status from active to permanently unusable status or temporarily unusable status. We used data from a large kidney care network from January 2021 to December 2021. Two models were created using LLMs and traditional TF-IDF technique. We used “distilbert-base-uncased”, a distilled version of BERT base model [1], and compared its performance with traditional TF-IDF-based NLP techniques. The dataset was randomly divided into 60% training, 20% validation and 20% test dataset. The test data, comprising of unseen patients’ data was used to evaluate the performance of the model. Both models were evaluated using metrics such as area under the receiver operating curve (AUROC), accuracy, sensitivity, and specificity. Results The incidence of 30 days AVF failure rate was 2.3% in the population. Both LLMs and traditional showed similar overall performance as summarized in Table 1. Notably, LLMs showed marginally better performance in certain evaluation metrics. Both models had same AUROC of 0.64 on test data. The accuracy and balanced accuracy for LLMs were 72.9% and 59.7%, respectively, compared to 70.9% and 59.6% for the traditional TF-IDF approach. In terms of specificity, LLMs scored 73.2%, slightly higher than the 71.2% observed for traditional NLP methods. However, LLMs had a lower sensitivity of 46.1% compared to 48% for traditional NLP. However, it is worth noting that training on LLMs took considerably longer than TF-IDF. Moreover, it also used higher computational resources such as utilization of graphics processing units (GPU) instances in cloud-based services, leading to higher cost. Conclusion In our study, we discovered that advanced LLMs perform comparably to traditional TF-IDF modeling techniques in predicting the failure of AVF. Both models demonstrated identical AUROC. While specificity was higher in LLMs compared to traditional NLP, sensitivity was higher in traditional NLP compared to LLMs. LLM was fine-tuned with a limited dataset, which could have influenced its performance to be similar to that of traditional NLP methods. This finding suggests that while LLMs may excel in certain scenarios, such as performing in-depth sentiment analysis of patient data for complex tasks, their effectiveness is highly dependent on the specific use case. It is crucial to weigh the benefits against the resources required for LLMs, as they can be significantly more resource-intensive and costly compared to traditional TF-IDF methods. This highlights the importance of a use-case-driven approach in selecting the appropriate NLP technique for healthcare applications.

On the role of the UMLS in supporting diagnosis generation proposed by Large Language Models

A novel prompting method for few-shot NER via LLMs

Brief hospital course (BHC) summaries are clinical documents that summarize a patient's hospital stay. While large language models (LLMs) depict remarkable capabilities in automating real-world tasks, their capabilities for healthcare applications such as synthesizing BHCs from clinical notes have not been shown. We introduce a novel preprocessed dataset, the MIMIC-IV-BHC, encapsulating clinical note and BHC pairs to adapt LLMs for BHC synthesis. Furthermore, we introduce a benchmark of the summarization performance of 2 general-purpose LLMs and 3 healthcare-adapted LLMs. Using clinical notes as input, we apply prompting-based (using in-context learning) and fine-tuning-based adaptation strategies to 3 open-source LLMs (Clinical-T5-Large, Llama2-13B, and FLAN-UL2) and 2 proprietary LLMs (Generative Pre-trained Transformer [GPT]-3.5 and GPT-4). We evaluate these LLMs across multiple context-length inputs using natural language similarity metrics. We further conduct a clinical study with 5 clinicians, comparing clinician-written and LLM-generated BHCs across 30 samples, focusing on their potential to enhance clinical decision-making through improved summary quality. We compare reader preferences for the original and LLM-generated summary using Wilcoxon signed-rank tests. We further request optional qualitative feedback from clinicians to gain deeper insights into their preferences, and we present the frequency of common themes arising from these comments. The Llama2-13B fine-tuned LLM outperforms other domain-adapted models given quantitative evaluation metrics of Bilingual Evaluation Understudy (BLEU) and Bidirectional Encoder Representations from Transformers (BERT)-Score. GPT-4 with in-context learning shows more robustness to increasing context lengths of clinical note inputs than fine-tuned Llama2-13B. Despite comparable quantitative metrics, the reader study depicts a significant preference for summaries generated by GPT-4 with in-context learning compared to both Llama2-13B fine-tuned summaries and the original summaries (P<.001), highlighting the need for qualitative clinical evaluation. We release a foundational clinically relevant dataset, the MIMIC-IV-BHC, and present an open-source benchmark of LLM performance in BHC synthesis from clinical notes. We observe high-quality summarization performance for both in-context proprietary and fine-tuned open-source LLMs using both quantitative metrics and a qualitative clinical reader study. Our research effectively integrates elements from the data assimilation pipeline: our methods use (1) clinical data sources to integrate, (2) data translation, and (3) knowledge creation, while our evaluation strategy paves the way for (4) deployment.

Generative large language models (LLMs) are a subset of transformers-based neural network architecture models. LLMs have successfully leveraged a combination of an increased number of parameters, improvements in computational efficiency, and large pre-training datasets to perform a wide spectrum of natural language processing (NLP) tasks. Using a few examples (few-shot) or no examples (zero-shot) for prompt-tuning has enabled LLMs to achieve state-of-the-art performance in a broad range of NLP applications. This article by the American Medical Informatics Association (AMIA) NLP Working Group characterizes the opportunities, challenges, and best practices for our community to leverage and advance the integration of LLMs in downstream NLP applications effectively. This can be accomplished through a variety of approaches, including augmented prompting, instruction prompt tuning, and reinforcement learning from human feedback (RLHF). Our focus is on making LLMs accessible to the broader biomedical informatics community, including clinicians and researchers who may be unfamiliar with NLP. Additionally, NLP practitioners may gain insight from the described best practices. We focus on 3 broad categories of NLP tasks, namely natural language understanding, natural language inferencing, and natural language generation. We review the emerging trends in prompt tuning, instruction fine-tuning, and evaluation metrics used for LLMs while drawing attention to several issues that impact biomedical NLP applications, including falsehoods in generated text (confabulation/hallucinations), toxicity, and dataset contamination leading to overfitting. We also review potential approaches to address some of these current challenges in LLMs, such as chain of thought prompting, and the phenomena of emergent capabilities observed in LLMs that can be leveraged to address complex NLP challenge in biomedical applications.

Large Language Models (LLMs) commonly report high confidence, even in domains where their underlying knowledge or training data is limited. This mismatch can negatively impact model reliability, particularly affecting educational applications where users may not recognize errors. To detect these knowledge gaps, LLM knowledge must be assessed after training. In this work, we compare LLM prompts to self-assess knowledge of content in two ways: rank-ordering and direct confidence scoring (e.g., 1-5). For human metacognition, rankings or A/B comparisons are more reliable, so we hypothesize that LLMs’ rankings may also be more effective than scores. We compare LLM-generated Overall Rankings and confidence scores for 15 topics against two external estimates of LLM knowledge: expert human ratings and search result counts from Bing, Google, and Wikipedia. We also consider Anchored Rankings in which each document to be rated by the LLM is compared to a set of documents with known expert scores. Comparing across different document representations and different LLMs, Spearman correlations with expert ratings are generally: positive and relatively high with Anchored Rankings having the highest correlation (ρ ranging from 0.74 to 1.0) followed by Overall Rankings and then confidence scores. In contrast, search-based signals have a weaker and variable alignment suggesting that web popularity is a noisy signal for estimating LLM familiarity with content. Overall, these findings suggest that relative self-assessment through rankings provides an interpretable signal of LLM self-knowledge. This can be used to select specialized prompts or workflows for topics where an LLM has less knowledge.

e13630 Background: As many as 60% of prior authorization requests are denied, yet coverage approval occurs for more than 60% of appeals for some therapies. Appeal processes encumber providers and increase burnout, but large language models (LLMs) may aid providers by drafting appeal letters. We evaluated LLM performance at this task for radiotherapy denials. Methods: Three commercially accessible LLMs were evaluated: generative pre-trained transformer 3.5 (GPT3.5), GPT4, and GPT4+web with internet search capacity (OpenAI, Inc., San Francisco, CA). A fourth LLM, GPT3.5-FT, was developed by fine-tuning GPT3.5 in a HIPAA-complaint local environment. The fine-tuning training data comprised 53 insurance denial appeal letters prepared by radiation oncologists and paired prompts describing the clinical history and appeal intent. Training data were enriched in appeal letters for proton radiotherapy, stereotactic body radiotherapy, and image-guided radiotherapy for myriad clinical scenarios. Twenty prompts, each requesting a letter for a simulated patient history, were programmatically presented to the LLMs. Three radiation oncologists, who were blinded to the LLM source, scored letter outputs across four domains: language syntax and semantics, clinical detail inclusion, clinical reasoning validity, and overall readiness for insurer submission. Additionally, one radiation oncologist scored the authenticity and relevance of literature sources cited in output letters, which were requested by several test prompts. Interobserver agreement between radiation oncologist scores was determined by Cohen’s kappa coefficient. Scores were compared between LLMs with non-parametric statistical tests. Results: Agreement between radiation oncologists’ scores was moderate-to-excellent across all domains (median κ = 0.68, minimum κ = 0.41). GPT3.5, GPT4, and GPT4+web drafted letters that, by mode average, were semantically and syntactically clear, included all provided clinical history without confabulation, clinically reasoned with few necessary revisions, and overall were submissible to an insurer with minor revisions. GPT4 and GPT4+web clinically reasoned better than GPT3.5 (p values &lt; 0.001). In contrast, GPT3.5-FT performance was inferior to other LLMs across all domains (p values &lt; 0.001). LLMs were poor at identifying, citing, and summarizing relevant literature unless provided in the prompt. Conclusions: LLMs can draft insurance appeal letters for radiotherapy services that require few revisions yet are poor at referencing relevant literature. Contrary to our hypothesis, fine-tuning with data from our department compromised LLM performance.

Recent works have shown that Large Language Models (LLMs) have a tendency to memorize patterns and biases present in their training data, raising important questions about how such memorized content influences model behavior.One such concern is the emergence of political bias in LLM outputs.In this paper, we investigate the extent to which LLMs' political leanings reflect memorized patterns from their pretraining corpora.We propose a method to quantitatively evaluate political leanings embedded in the large pretraining corpora.Subsequently we investigate to whom are the LLMs' political leanings more aligned with, their pretrainig corpora or the surveyed human opinions.As a case study, we focus on probing the political leanings of LLMs in 32 U.S. Supreme Court cases, addressing contentious topics such as abortion and voting rights.Our findings reveal that LLMs strongly reflect the political leanings in their training data, and no strong correlation is observed with their alignment to human opinions as expressed in surveys.These results underscore the importance of responsible curation of training data, and the methodology for auditing the memorization in LLMs to ensure human-AI alignment.

Model tuning or prompt Tuning? a study of large language models for clinical concept and relation extraction

Custom Large Language Models Improve Accuracy: Comparing Retrieval Augmented Generation and Artificial Intelligence Agents to Non-Custom Models for Evidence-Based Medicine

BackgroundThe launch of ChatGPT (OpenAI) in November 2022 attracted public attention and academic interest to large language models (LLMs), facilitating the emergence of many other innovative LLMs. These LLMs have been applied in various fields, including health care. Numerous studies have since been conducted regarding how to use state-of-the-art LLMs in health-related scenarios.ObjectiveThis review aims to summarize applications of and concerns regarding conversational LLMs in health care and provide an agenda for future research in this field.MethodsWe used PubMed, ACM, and the IEEE digital libraries as primary sources for this review. We followed the guidance of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) to screen and select peer-reviewed research articles that (1) were related to health care applications and conversational LLMs and (2) were published before September 1, 2023, the date when we started paper collection. We investigated these papers and classified them according to their applications and concerns.ResultsOur search initially identified 820 papers according to targeted keywords, out of which 65 (7.9%) papers met our criteria and were included in the review. The most popular conversational LLM was ChatGPT (60/65, 92% of papers), followed by Bard (Google LLC; 1/65, 2% of papers), LLaMA (Meta; 1/65, 2% of papers), and other LLMs (6/65, 9% papers). These papers were classified into four categories of applications: (1) summarization, (2) medical knowledge inquiry, (3) prediction (eg, diagnosis, treatment recommendation, and drug synergy), and (4) administration (eg, documentation and information collection), and four categories of concerns: (1) reliability (eg, training data quality, accuracy, interpretability, and consistency in responses), (2) bias, (3) privacy, and (4) public acceptability. There were 49 (75%) papers using LLMs for either summarization or medical knowledge inquiry, or both, and there are 58 (89%) papers expressing concerns about either reliability or bias, or both. We found that conversational LLMs exhibited promising results in summarization and providing general medical knowledge to patients with a relatively high accuracy. However, conversational LLMs such as ChatGPT are not always able to provide reliable answers to complex health-related tasks (eg, diagnosis) that require specialized domain expertise. While bias or privacy issues are often noted as concerns, no experiments in our reviewed papers thoughtfully examined how conversational LLMs lead to these issues in health care research.ConclusionsFuture studies should focus on improving the reliability of LLM applications in complex health-related tasks, as well as investigating the mechanisms of how LLM applications bring bias and privacy issues. Considering the vast accessibility of LLMs, legal, social, and technical efforts are all needed to address concerns about LLMs to promote, improve, and regularize the application of LLMs in health care.

Despite the impressive performance across numerous tasks, Large Language Models (LLMs) often fail in solving simple decision-making tasks due to the misalignment of the knowledge in LLMs with environments. On the contrary, Reinforcement Learning (RL) agents learn policies from scratch, which makes them always align with environments but difficult to incorporate prior knowledge for efficient explorations. To narrow the gap, we propose TWOSOME, a novel general online framework that deploys LLMs as decision-making agents to efficiently interact and align with embodied environments via RL without requiring any prepared datasets or prior knowledge of the environments. First, we query the joint probabilities of each valid action with LLMs to form behavior policies. Then, to enhance the stability and robustness of the policies, we propose two normalization methods and summarize four prompt design principles. Finally, we design a novel parameter-efficient training architecture where the actor and critic share one frozen LLM equipped with LOw-Rank Adapters (LoRA) updated by PPO. We conduct extensive experiments to evaluate TWOSOME. (i) TWOSOME exhibits significantly better sample efficiency and performance compared to the conventional RL method, PPO, and prompt tuning method, SayCan, in both classical decision-making environment, Overcooked, and simulated household environment, VirtualHome. (ii) Benefiting from LLMs’ open-vocabulary feature, TWOSOME shows superior generalization ability to unseen tasks. (iii) Under our framework, there is no significant loss of the LLMs’ original ability during online PPO finetuning.a

Background: Recent advancements in large language model (LLM) technologies have introduced powerful open-source instruction-tuned LLMs that match the text generation quality of leading models like GPT-4. Despite accelerating LLM adoption in sensitive-information environments, the lack of disclosed training data hinders replication and makes these achievements exclusive to specific models. Purpose: Given the multilingual nature of the latest iteration of open-source LLMs, the benefits of training language-specific LLMs diminish, leaving computational efficiency as the sole guaranteed advantage of this computationally-expensive procedure. This work aims to address the language-adaptation limitations posed by restricted access to high-quality instruction-tuning data, offering a more cost-effective pipeline. Method: To tackle language-adaptation challenges, we introduce Learned Embedding Propagation (LEP), a novel method with lower training data requirements and minimal disruption of existing LLM knowledge. LEP employs an innovative embedding propagation technique, bypassing the need for instruction-tuning and directly integrating new language knowledge into any instruct-tuned LLM variant. Additionally, we developed Darumeru, a new benchmark for evaluating text generation robustness during training, specifically tailored for Russian adaptation. Results: We applied the LEP method to adapt LLaMa-3-8B and Mistral-7B for Russian, testing four different vocabulary adaptation scenarios. Evaluation demonstrates that LEP achieves competitive performance levels, comparable to OpenChat 3.5 and LLaMa-3-8B-Instruct. Further improvements were observed through self-calibration and additional instruction-tuning steps, enhancing task-solving capabilities beyond the original models. Conclusion: LEP offers a viable and efficient alternative to traditional language-specific instruction-tuning, significantly reducing the costs associated with language adaptation while maintaining or surpassing the performance benchmarks set by contemporary LLMs.

Cognitive biases are systematic deviations in thinking that lead to irrational judgments and problematic decision-making, extensively studied across various fields. Recently, large language models (LLMs) have shown advanced understanding capabilities but may inherit human biases from their training data. While social biases in LLMs have been well-studied, cognitive biases have received less attention, with existing research focusing on specific scenarios. The broader impact of cognitive biases on LLMs in various decision-making contexts remains underexplored. We investigated whether LLMs are influenced by the threshold priming effect in relevance judgments, a core task and widely-discussed research topic in the Information Retrieval (IR) coummunity. The priming effect occurs when exposure to certain stimuli unconsciously affects subsequent behavior and decisions. Our experiment employed 10 topics from the TREC 2019 Deep Learning passage track collection, and tested AI judgments under different document relevance scores, batch lengths, and LLM models, including GPT-3.5, GPT-4, LLaMa2-13B and LLaMa2-70B. Results showed that LLMs tend to give lower scores to later documents if earlier ones have high relevance, and vice versa, regardless of the combination and model used. Our finding demonstrates that LLM's judgments, similar to human judgments, are also influenced by threshold priming biases, and suggests that researchers and system engineers should take into account potential human-like cognitive biases in designing, evaluating, and auditing LLMs in IR tasks and beyond

Background:Even though patients have easy access to their electronic health records and lab test results data through patient portals, lab results are often confusing and hard to understand. Many patients turn to online forums or question and answering (Q&A) sites to seek advice from their peers. However, the quality of answers from social Q&A on health-related questions varies significantly, and not all the responses are accurate or reliable. Large language models (LLMs) such as ChatGPT have opened a promising avenue for patients to get their questions answered.Objective:We aim to assess the feasibility of using LLMs to generate relevant, accurate, helpful, and unharmful responses to lab test-related questions asked by patients and to identify potential issues that can be mitigated with augmentation approaches.Methods:We first collected lab test results related question and answer data from Yahoo! Answers and selected 53 Q&A pairs for this study. Using the LangChain framework and ChatGPT web portal, we generated responses to the 53 questions from four LLMs including GPT-4, Meta LLaMA 2, MedAlpaca, and ORCA_mini. We first assessed the similarity of their answers using standard QA similarity-based evaluation metrics including ROUGE, BLEU, METEOR, BERTScore. We also utilized an LLM-based evaluator to judge whether a target model has higher quality in terms of relevance, correctness, helpfulness, and safety than the baseline model. Finally, we performed a manual evaluation with medical experts for all the responses of seven selected questions on the same four aspects.Results:Regarding the similarity of the responses from 4 LLMs, where GPT-4 output was used as the reference answer, the responses from LLaMa 2 are the most similar ones, followed by LLaMa 2, ORCA_mini, and MedAlpaca. Human answers from Yahoo data were scored lowest and thus least similar to GPT-4-generated answers. The results of Win Rate and medical expert evaluation both showed that GPT-4’s responses achieved better scores than all the other LLM responses and human responses on all the four aspects (relevance, correctness, helpfulness, and safety). However, LLM responses occasionally also suffer from lack of interpretation in one’s medical context, incorrect statements, and lack of references.Conclusions:By evaluating LLMs in generating responses to patients’ lab test results related questions, we find that compared to other three LLMs and human answer from the Q&A website, GPT-4’s responses are more accurate, helpful, relevant, and safer. However, there are cases that GPT-4 responses are inaccurate and not individualized. We identified a number of ways to improve the quality of LLM responses including prompt engineering, prompt augmentation, retrieval augmented generation, and response evaluation.