Dataset Information

Development of a Human Evaluation Framework and Correlation with Automated Metrics for Natural Language Generation of Medical Diagnoses.

ABSTRACT: In the evolving landscape of clinical Natural Language Generation (NLG), assessing abstractive text quality remains challenging, as existing methods often overlook generative task complexities. This work aimed to examine the current state of automated evaluation metrics in NLG in healthcare. To have a robust and well-validated baseline with which to examine the alignment of these metrics, we created a comprehensive human evaluation framework. Employing ChatGPT-3.5-turbo generative output, we correlated human judgments with each metric. None of the metrics demonstrated high alignment; however, the SapBERT score-a Unified Medical Language System (UMLS)- showed the best results. This underscores the importance of incorporating domain-specific knowledge into evaluation efforts. Our work reveals the deficiency in quality evaluations for generated text and introduces our comprehensive human evaluation framework as a baseline. Future efforts should prioritize integrating medical knowledge databases to enhance the alignment of automated metrics, particularly focusing on refining the SapBERT score for improved assessments.

SUBMITTER: Croxford E

PROVIDER: S-EPMC10984060 | biostudies-literature | 2024 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Development of a Human Evaluation Framework and Correlation with Automated Metrics for Natural Language Generation of Medical Diagnoses.

Croxford Emma E Gao Yanjun Y Patterson Brian B To Daniel D Tesch Samuel S Dligach Dmitriy D Mayampurath Anoop A Churpek Matthew M MM Afshar Majid M

medRxiv : the preprint server for health sciences 20240409

In the evolving landscape of clinical Natural Language Generation (NLG), assessing abstractive text quality remains challenging, as existing methods often overlook generative task complexities. This work aimed to examine the current state of automated evaluation metrics in NLG in healthcare. To have a robust and well-validated baseline with which to examine the alignment of these metrics, we created a comprehensive human evaluation framework. Employing ChatGPT-3.5-turbo generative output, we cor ...[more]

PMID: 38562730

Similar Datasets

Project description:BackgroundMedical texts present significant domain-specific challenges, and manually curating these texts is a time-consuming and labor-intensive process. To address this, natural language processing (NLP) algorithms have been developed to automate text processing. In the biomedical field, various toolkits for text processing exist, which have greatly improved the efficiency of handling unstructured text. However, these existing toolkits tend to emphasize different perspectives, and none of them offer generation capabilities, leaving a significant gap in the current offerings.ObjectiveThis study aims to describe the development and preliminary evaluation of Ascle. Ascle is tailored for biomedical researchers and clinical staff with an easy-to-use, all-in-one solution that requires minimal programming expertise. For the first time, Ascle provides 4 advanced and challenging generative functions: question-answering, text summarization, text simplification, and machine translation. In addition, Ascle integrates 12 essential NLP functions, along with query and search capabilities for clinical databases.MethodsWe fine-tuned 32 domain-specific language models and evaluated them thoroughly on 27 established benchmarks. In addition, for the question-answering task, we developed a retrieval-augmented generation (RAG) framework for large language models that incorporated a medical knowledge graph with ranking techniques to enhance the reliability of generated answers. Additionally, we conducted a physician validation to assess the quality of generated content beyond automated metrics.ResultsThe fine-tuned models and RAG framework consistently enhanced text generation tasks. For example, the fine-tuned models improved the machine translation task by 20.27 in terms of BLEU score. In the question-answering task, the RAG framework raised the ROUGE-L score by 18% over the vanilla models. Physician validation of generated answers showed high scores for readability (4.95/5) and relevancy (4.43/5), with a lower score for accuracy (3.90/5) and completeness (3.31/5).ConclusionsThis study introduces the development and evaluation of Ascle, a user-friendly NLP toolkit designed for medical text generation. All code is publicly available through the Ascle GitHub repository. All fine-tuned language models can be accessed through Hugging Face.

Project description:BackgroundThe advancement of large language models (LLMs) offers significant opportunities for health care, particularly in the generation of medical documentation. However, challenges related to ensuring the accuracy and reliability of LLM outputs, coupled with the absence of established quality standards, have raised concerns about their clinical application.ObjectiveThis study aimed to develop and validate an evaluation framework for assessing the accuracy and clinical applicability of LLM-generated emergency department (ED) records, aiming to enhance artificial intelligence integration in health care documentation.MethodsWe organized the Healthcare Prompt-a-thon, a competitive event designed to explore the capabilities of LLMs in generating accurate medical records. The event involved 52 participants who generated 33 initial ED records using HyperCLOVA X, a Korean-specialized LLM. We applied a dual evaluation approach. First, clinical evaluation: 4 medical professionals evaluated the records using a 5-point Likert scale across 5 criteria-appropriateness, accuracy, structure/format, conciseness, and clinical validity. Second, quantitative evaluation: We developed a framework to categorize and count errors in the LLM outputs, identifying 7 key error types. Statistical methods, including Pearson correlation and intraclass correlation coefficients (ICC), were used to assess consistency and agreement among evaluators.ResultsThe clinical evaluation demonstrated strong interrater reliability, with ICC values ranging from 0.653 to 0.887 (P<.001), and a test-retest reliability Pearson correlation coefficient of 0.776 (P<.001). Quantitative analysis revealed that invalid generation errors were the most common, constituting 35.38% of total errors, while structural malformation errors had the most significant negative impact on the clinical evaluation score (Pearson r=-0.654; P<.001). A strong negative correlation was found between the number of quantitative errors and clinical evaluation scores (Pearson r=-0.633; P<.001), indicating that higher error rates corresponded to lower clinical acceptability.ConclusionsOur research provides robust support for the reliability and clinical acceptability of the proposed evaluation framework. It underscores the framework's potential to mitigate clinical burdens and foster the responsible integration of artificial intelligence technologies in health care, suggesting a promising direction for future research and practical applications in the field.

Project description:BackgroundManually extracted data points from health records are collated on an institutional, provincial, and national level to facilitate clinical research. However, the labour-intensive clinical chart review process puts an increasing burden on healthcare system budgets. Therefore, an automated information extraction system is needed to ensure the timeliness and scalability of research data.MethodsWe used a dataset of 100 synoptic operative and 100 pathology reports, evenly split into 50 reports in training and test sets for each report type. The training set guided our development of a Natural Language Processing (NLP) extraction pipeline system, which accepts scanned images of operative and pathology reports. The system uses a combination of rule-based and transfer learning methods to extract numeric encodings from text. We also developed visualization tools to compare the manual and automated extractions. The code for this paper was made available on GitHub.ResultsA test set of 50 operative and 50 pathology reports were used to evaluate the extraction accuracies of the NLP pipeline. Gold standard, defined as manual extraction by expert reviewers, yielded accuracies of 90.5% for operative reports and 96.0% for pathology reports, while the NLP system achieved overall 91.9% (operative) and 95.4% (pathology) accuracy. The pipeline successfully extracted outcomes data pertinent to breast cancer tumor characteristics (e.g. presence of invasive carcinoma, size, histologic type), prognostic factors (e.g. number of lymph nodes with micro-metastases and macro-metastases, pathologic stage), and treatment-related variables (e.g. margins, neo-adjuvant treatment, surgical indication) with high accuracy. Out of the 48 variables across operative and pathology codebooks, NLP yielded 43 variables with F-scores of at least 0.90; in comparison, a trained human annotator yielded 44 variables with F-scores of at least 0.90.ConclusionsThe NLP system achieves near-human-level accuracy in both operative and pathology reports using a minimal curated dataset. This system uniquely provides a robust solution for transparent, adaptable, and scalable automation of data extraction from patient health records. It may serve to advance breast cancer clinical research by facilitating collection of vast amounts of valuable health data at a population level.

Dataset Information

Development of a Human Evaluation Framework and Correlation with Automated Metrics for Natural Language Generation of Medical Diagnoses.

Publications

Development of a Human Evaluation Framework and Correlation with Automated Metrics for Natural Language Generation of Medical Diagnoses.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets