Aci-bench: a Novel Ambient Clinical Intelligence Dataset for Benchmarking Automatic Visit Note Generation

Artificial intelligence-generated feedback on social signals in patient-provider communication: technical performance, feedback usability, and impact

Objectives

Implicit bias perpetuates health care inequities and manifests in patient-provider interactions, particularly nonverbal social cues like dominance. We investigated the use of artificial intelligence (AI) for automated communication assessment and feedback during primary care visits to raise clinician awareness of bias in patient interactions.

Materials and Methods

(1) Assessed the technical performance of our AI models by building a machine-learning pipeline that automatically detects social signals in patient-provider interactions from 145 primary care visits. (2) Engaged 24 clinicians to design usable AI-generated communication feedback for their workflow. (3) Evaluated the impact of our AI-based approach in a prospective cohort of 108 primary care visits.

Results

Findings demonstrate the feasibility of AI models to identify social signals, such as dominance, warmth, engagement, and interactivity, in nonverbal patient-provider communication. Although engaged clinicians preferred feedback delivered in personalized dashboards, they found nonverbal cues difficult to interpret, motivating social signals as an alternative feedback mechanism. Impact evaluation demonstrated fairness in all AI models with better generalizability of provider dominance, provider engagement, and patient warmth. Stronger clinician implicit race bias was associated with less provider dominance and warmth. Although clinicians expressed overall interest in our AI approach, they recommended improvements to enhance acceptability, feasibility, and implementation in telehealth and medical education contexts.

Discussion and Conclusion

Findings demonstrate promise for AI-driven communication assessment and feedback systems focused on social signals. Future work should improve the performance of this approach, personalize models, and contextualize feedback, and investigate system implementation in educational workflows. This work exemplifies a systematic, multistage approach for evaluating AI tools designed to raise clinician awareness of implicit bias and promote patient-centered, equitable health care interactions.

Promises and pitfalls of artificial intelligence models in forecasting rheumatoid arthritis treatment response and outcomes

None.

CACER: Clinical concept Annotations for Cancer Events and Relations

OBJECTIVE

Clinical notes contain unstructured representations of patient histories, including the relationships between medical problems and prescription drugs. To investigate the relationship between cancer drugs and their associated symptom burden, we extract structured, semantic representations of medical problem and drug information from the clinical narratives of oncology notes.

MATERIALS AND METHODS

We present Clinical concept Annotations for Cancer Events and Relations (CACER), a novel corpus with fine-grained annotations for over 48 000 medical problems and drug events and 10 000 drug-problem and problem-problem relations. Leveraging CACER, we develop and evaluate transformer-based information extraction models such as Bidirectional Encoder Representations from Transformers (BERT), Fine-tuned Language Net Text-To-Text Transfer Transformer (Flan-T5), Large Language Model Meta AI (Llama3), and Generative Pre-trained Transformers-4 (GPT-4) using fine-tuning and in-context learning (ICL).

RESULTS

In event extraction, the fine-tuned BERT and Llama3 models achieved the highest performance at 88.2-88.0 F1, which is comparable to the inter-annotator agreement (IAA) of 88.4 F1. In relation extraction, the fine-tuned BERT, Flan-T5, and Llama3 achieved the highest performance at 61.8-65.3 F1. GPT-4 with ICL achieved the worst performance across both tasks.

DISCUSSION

The fine-tuned models significantly outperformed GPT-4 in ICL, highlighting the importance of annotated training data and model optimization. Furthermore, the BERT models performed similarly to Llama3. For our task, large language models offer no performance advantage over the smaller BERT models.

CONCLUSIONS

We introduce CACER, a novel corpus with fine-grained annotations for medical problems, drugs, and their relationships in clinical narratives of oncology notes. State-of-the-art transformer models achieved performance comparable to IAA for several extraction tasks.

Testing and Evaluation of Health Care Applications of Large Language Models: A Systematic Review

Importance

Large language models (LLMs) can assist in various health care activities, but current evaluation approaches may not adequately identify the most useful application areas.

Objective

To summarize existing evaluations of LLMs in health care in terms of 5 components: (1) evaluation data type, (2) health care task, (3) natural language processing (NLP) and natural language understanding (NLU) tasks, (4) dimension of evaluation, and (5) medical specialty.

Data Sources

A systematic search of PubMed and Web of Science was performed for studies published between January 1, 2022, and February 19, 2024.

Study Selection

Studies evaluating 1 or more LLMs in health care.

Data Extraction and Synthesis

Three independent reviewers categorized studies via keyword searches based on the data used, the health care tasks, the NLP and NLU tasks, the dimensions of evaluation, and the medical specialty.

Results

Of 519 studies reviewed, published between January 1, 2022, and February 19, 2024, only 5% used real patient care data for LLM evaluation. The most common health care tasks were assessing medical knowledge such as answering medical licensing examination questions (44.5%) and making diagnoses (19.5%). Administrative tasks such as assigning billing codes (0.2%) and writing prescriptions (0.2%) were less studied. For NLP and NLU tasks, most studies focused on question answering (84.2%), while tasks such as summarization (8.9%) and conversational dialogue (3.3%) were infrequent. Almost all studies (95.4%) used accuracy as the primary dimension of evaluation; fairness, bias, and toxicity (15.8%), deployment considerations (4.6%), and calibration and uncertainty (1.2%) were infrequently measured. Finally, in terms of medical specialty area, most studies were in generic health care applications (25.6%), internal medicine (16.4%), surgery (11.4%), and ophthalmology (6.9%), with nuclear medicine (0.6%), physical medicine (0.4%), and medical genetics (0.2%) being the least represented.

Conclusions and Relevance

Existing evaluations of LLMs mostly focus on accuracy of question answering for medical examinations, without consideration of real patient care data. Dimensions such as fairness, bias, and toxicity and deployment considerations received limited attention. Future evaluations should adopt standardized applications and metrics, use clinical data, and broaden focus to include a wider range of tasks and specialties.

Exploring the Feasibility of GPT-4 as a Data Extraction Tool for Renal Surgery Operative Notes

INTRODUCTION

GPT-4 is a large language model with potential for multiple applications in urology. Our study sought to evaluate GPT-4's performance in data extraction from renal surgery operative notes.

METHODS

GPT-4 was queried to extract information on laterality, surgery, approach, estimated blood loss, and ischemia time from deidentified operative notes. Match rates were determined by the number of "matched" data points between GPT-4 and human-curated extraction. Accuracy rates were calculated after manually reviewing "not matched" data points. Cohen's kappa and the intraclass coefficient were used to evaluate interrater agreement/reliability.

RESULTS

Our cohort consisted of 1498 renal surgeries from 2003 to 2023. Match rates were high for laterality (94.4%), surgery (92.5%), and approach (89.4%), but lower for estimated blood loss (77.1%) and ischemia time (25.6%). GPT-4 was more accurate for estimated blood loss (90.3% vs 85.5% human curated) and similarly accurate for laterality (95.2% vs 95.3% human curated). Human-curated accuracy rates were higher for surgery (99.3% vs 93% GPT-4), approach (97.9% vs 90.8% GPT-4), and ischemia time (95.6% vs 30.7% GPT-4). Cohen's kappa was 0.96 for laterality, 0.83 for approach, and 0.71 for surgery. The intraclass coefficient was 0.62 for estimated blood loss and 0.09 for ischemia time.

CONCLUSIONS

Match and accuracy rates were higher for categorical variables. GPT-4 data extraction was particularly error prone for variables with heterogenous documentation styles. The role of a standard operative template to aid data extraction will be explored in the future. GPT-4 can be utilized as a helpful and efficient data extraction tool with manual feedback.

Adapted large language models can outperform medical experts in clinical text summarization

Analyzing vast textual data and summarizing key information from electronic health records imposes a substantial burden on how clinicians allocate their time. Although large language models (LLMs) have shown promise in natural language processing (NLP) tasks, their effectiveness on a diverse range of clinical summarization tasks remains unproven. Here we applied adaptation methods to eight LLMs, spanning four distinct clinical summarization tasks: radiology reports, patient questions, progress notes and doctor-patient dialogue. Quantitative assessments with syntactic, semantic and conceptual NLP metrics reveal trade-offs between models and adaptation methods. A clinical reader study with 10 physicians evaluated summary completeness, correctness and conciseness; in most cases, summaries from our best-adapted LLMs were deemed either equivalent (45%) or superior (36%) compared with summaries from medical experts. The ensuing safety analysis highlights challenges faced by both LLMs and medical experts, as we connect errors to potential medical harm and categorize types of fabricated information. Our research provides evidence of LLMs outperforming medical experts in clinical text summarization across multiple tasks. This suggests that integrating LLMs into clinical workflows could alleviate documentation burden, allowing clinicians to focus more on patient care.