Medical Information Extraction Research Articles

Clinical information extraction for lower-resource languages and domains with few-shot learning using pretrained language models and prompting

Abstract A vast amount of clinical data are still stored in unstructured text. Automatic extraction of medical information from these data poses several challenges: high costs of clinical expertise, restricted computational resources, strict privacy regulations, and limited interpretability of model predictions. Recent domain adaptation and prompting methods using lightweight masked language models showed promising results with minimal training data and allow for application of well-established interpretability methods. We are first to present a systematic evaluation of advanced domain-adaptation and prompting methods in a lower-resource medical domain task, performing multi-class section classification on German doctor’s letters. We evaluate a variety of models, model sizes (further-pre)training and task settings, and conduct extensive class-wise evaluations supported by Shapley values to validate the quality of small-scale training data and to ensure interpretability of model predictions. We show that in few-shot learning scenarios, a lightweight, domain-adapted pretrained language model, prompted with just 20 shots per section class, outperforms a traditional classification model, by increasing accuracy from $48.6\%$ to $79.1\%$ . By using Shapley values for model selection and training data optimization, we could further increase accuracy up to $84.3\%$ . Our analyses reveal that pretraining of masked language models on general-language data is important to support successful domain-transfer to medical language, so that further-pretraining of general-language models on domain-specific documents can outperform models pretrained on domain-specific data only. Our evaluations show that applying prompting based on general-language pretrained masked language models combined with further-pretraining on medical-domain data achieves significant improvements in accuracy beyond traditional models with minimal training data. Further performance improvements and interpretability of results can be achieved, using interpretability methods such as Shapley values. Our findings highlight the feasibility of deploying powerful machine learning methods in clinical settings and can serve as a process-oriented guideline for lower-resource languages and domains such as clinical information extraction projects.

MedT2T: An adaptive pointer constrain generating method for a new medical text-to-table task

Medical information extraction is a crucial task in the governance of healthcare data within medical information systems in the medical internet network, aimed at extracting vital information from existing content. However, structuring this key information into a table is currently a challenge, hindering the development of AI-driven smart health. In this study, we study the medical text-to-table task based on a new generative perspective. To address the challenges of ineffective numerical embedding, flexible table formats, and dense medical terminology and numerical entities in an end-to-end manner, we present the innovative medical text-to-table model called MedT2T. This model, built on the BART backbone, operates in an end-to-end manner and comprises three essential modules: Encoder, Decoder, and Adapter. The Encoder utilizes an innovative adaptive medical numerical constraint to facilitate precise embedding and generation of medical numerical data. The generated output of the Decoder adheres to relational constraints and table formats, ensuring the desired structure and organization. Additionally, the Adapter incorporates an adaptive pointer generation mechanism, allowing for dynamic referencing of medical terminology and numerical information either from the source text or generated through the vocabulary distribution of the Decoder. Our method outperforms existing baselines in terms of exact match, character level match, and BERTScore. We also proved that MedT2T can serve as an essential table extraction tool to bring informative gains for medical downstream classifiers and predictors. This study not only achieved accurate entity generation for tables from lengthy medical texts to improve physician efficiency in accessing critical information for decision-making, but also provided large-scale structured training table data for downstream tasks such as AI-driven smart healthcare.

Integrating AI with MBSE for Data Extraction from Medical Standards

AbstractThe growing adoption of Model‐Based Systems Engineering (MBSE) in the medical sector has prompted a significant emphasis on the digitization of medical standards into norm models. This transformation promotes consistency and allows for tracing system model elements to the corresponding norm model elements. Despite these efforts, the current digitization activities heavily rely on manual extraction and transformation, particularly from PDF documents into SysML models. Concurrently, the proliferation of Artificial Intelligence (AI) applications in recent years presents an opportunity to automate such activities. This paper contributes to the integration of AI with MBSE, focusing solely on the extraction and transformation of medical standards information from documents into SysML norm models. It explores the initial outcomes of augmenting data extraction from medical standards using recent AI algorithms and integrating them into MBSE practices. The evaluation involves two approaches, an open‐source multimodal classifier model and a proprietary large language model. The study assesses these approaches on a medical standard and outlines future work, including the exploration of an open‐source large language model approach.

The faster or richer the response, the better performance? An empirical analysis of online healthcare platforms from a competitive perspective

The emergence of online healthcare platforms has changed the competitive environment among physicians. However, little is known about how physicians can improve their performance in this new environment. Platforms also face challenges in comprehending the competitive mechanisms among physicians, which might hinder them from formulating strategic managerial decisions that foster sustained growth. In this light, we extract medical service-related information from physicians' response behavioral data on a prominent healthcare platform, and empirically investigate the factors affecting physicians' online performance from a competitive perspective as well as the gender differences in these effects. The results indicate that physicians' responses significantly impact their online performance, revealing a competitive relationship between physicians and their colleagues in the same department. Specifically, a fast response time and informative responses are positively correlated with the focal physician's performance, whereas colleagues' informative responses negatively impact the focal physician's performance, and this relationship is mediated by the focal physician's response informativeness. Nevertheless, there is no significant correlation between colleagues' response time and the focal physician's performance. The results also unveil that gender moderates the effect of response informativeness on the focal physician's performance. Specifically, colleagues' response informativeness has a more significant impact on male physicians' performance than on female physicians' performance, suggesting a greater propensity for competition among male physicians. Our findings could offer decision support for enhancing physician performance and healthcare platform management.

Using the Natural Language Processing System Medical Named Entity Recognition-Japanese to Analyze Pharmaceutical Care Records: Natural Language Processing Analysis.

Large language models have propelled recent advances in artificial intelligence technology, facilitating the extraction of medical information from unstructured data such as medical records. Although named entity recognition (NER) is used to extract data from physicians' records, it has yet to be widely applied to pharmaceutical care records. In this study, we aimed to investigate the feasibility of automatic extraction of the information regarding patients' diseases and symptoms from pharmaceutical care records. The verification was performed using Medical Named Entity Recognition-Japanese (MedNER-J), a Japanese disease-extraction system designed for physicians' records. MedNER-J was applied to subjective, objective, assessment, and plan data from the care records of 49 patients who received cefazolin sodium injection at Keio University Hospital between April 2018 and March 2019. The performance of MedNER-J was evaluated in terms of precision, recall, and F1-score. The F1-scores of NER for subjective, objective, assessment, and plan data were 0.46, 0.70, 0.76, and 0.35, respectively. In NER and positive-negative classification, the F1-scores were 0.28, 0.39, 0.64, and 0.077, respectively. The F1-scores of NER for objective (0.70) and assessment data (0.76) were higher than those for subjective and plan data, which supported the superiority of NER performance for objective and assessment data. This might be because objective and assessment data contained many technical terms, similar to the training data for MedNER-J. Meanwhile, the F1-score of NER and positive-negative classification was high for assessment data alone (F1-score=0.64), which was attributed to the similarity of its description format and contents to those of the training data. MedNER-J successfully read pharmaceutical care records and showed the best performance for assessment data. However, challenges remain in analyzing records other than assessment data. Therefore, it will be necessary to reinforce the training data for subjective data in order to apply the system to pharmaceutical care records.

Building Trustworthy Generative Artificial Intelligence for Diabetes Care and Limb Preservation: A Medical Knowledge Extraction Case.

Large language models (LLMs) offer significant potential in medical information extraction but carry risks of generating incorrect information. This study aims to develop and validate a retriever-augmented generation (RAG) model that provides accurate medical knowledge about diabetes and diabetic foot care to laypersons with an eighth-grade literacy level. Improving health literacy through patient education is paramount to addressing the problem of limb loss in the diabetic population. In addition to affecting patient well-being through improved outcomes, improved physician well-being is an important outcome of a self-management model for patient health education. We used an RAG architecture and built a question-and-answer artificial intelligence (AI) model to extract knowledge in response to questions pertaining to diabetes and diabetic foot care. We utilized GPT-4 by OpenAI, with Pinecone as a vector database. The NIH National Standards for Diabetes Self-Management Education served as the basis for our knowledge base. The model's outputs were validated through expert review against established guidelines and literature. Fifty-eight keywords were used to select 295 articles and the model was tested against 175 questions across topics. The study demonstrated that with appropriate content volume and few-shot learning prompts, the RAG model achieved 98% accuracy, confirming its capability to offer user-friendly and comprehensible medical information. The RAG model represents a promising tool for delivering reliable medical knowledge to the public which can be used for self-education and self-management for diabetes, highlighting the importance of content validation and innovative prompt engineering in AI applications.

PRTA:Joint extraction of medical nested entities and overlapping relation via parameter sharing progressive recognition and targeted assignment decoding scheme

Nested entities and relationship extraction are two tasks for analysis of electronic medical records. However, most of existing medical information extraction models consider these tasks separately, resulting in a lack of consistency between them. In this paper, we propose a joint medical entity-relation extraction model with progressive recognition and targeted assignment (PRTA). Entities and relations share the information of sequence and word embedding layers in the joint decoding stage. They are trained simultaneously and realize information interaction by updating the shared parameters. Specifically, we design a compound triangle strategy for the nested entity recognition and an adaptive multi-space interactive strategy for relationship extraction. Then, we construct a parameter-shared information space based on semantic continuity to decode entities and relationships. Extensive experiments were conducted on the Private Liver Disease Dataset (PLDD) provided by Beijing Friendship Hospital of Capital Medical University and public datasets (NYT, ACE04 and ACE05). The results show that our method outperforms existing SOTA methods in most indicators, and effectively handles nested entities and overlapping relationships.

Chinese medical short text classification model based on DPECNN

Medical short text classification is of great significance to medical information extraction and medical auxiliary diagnosis. However, medical short texts face challenges such as sparse features, semantic ambiguity, and the specialized nature of the medical field, resulting in relatively low accuracy in short text classification. Taking into consideration the characteristics of medical short texts, this paper proposes a Chinese medical short text classification model based on DPECNN. First, ERNIE is utilized to learn text knowledge and information in order to enhance the model’s semantic representation capabilities. Then, the DPECNN model is employed to extract rich feature information, and the classification results are generated through a fully connected layer. In the case of DPCNN, it only considers deep-level contextual semantic information, overlooking the correlation of adjacent semantic information between channels. To address this, ECA channel attention is introduced to account for adjacent semantic information. The use of a self-normalizing activation function helps avoid the problem of vanishing gradients. To enhance the model’s robustness and generalization ability, the FGM adversarial training algorithm is employed to perturb the data. The F1 values achieved on the THUCNews, KUAKE-QIC, and CHIP-CTC datasets are 95.00%, 79.45%, and 82.81%, respectively.

Intelligent extraction of medical entity relationship based on graph neural network and optimization strategy

Machine Learning technologies have obtained breakthrough in various fields, and medical information extraction has also successfully made great progress through deep learning methods. However, entity relationship overlap is a key issue and challenge in the field of medical entity relationship extraction. Hence, we propose an intelligent extraction method for Chinese medicine entity relationship through improving graph neural network and structure optimization strategy. We optimize model structure and introduce global pointer network. The sentence feature information is captured by attention mechanism and Bi-Long Short Term Memory (LSTM). Grammatical relations between entities are parsed using Graph Convolution Network (GCN) layer. Finally, multiple decoders are used for joint extraction of entity relations through the global pointer network. The model learns the graph structure data by GCN, while the challenge of entity relationship overlap is solved through using the pointer network. The experiments demonstrate that the model achieves 83.77% in terms of F1 value, which is better than other baseline models.

MED-Prompt: A novel prompt engineering framework for medicine prediction on free-text clinical notes

Existing AI-based medicine prediction systems require substantial training time, computing resources, and extensive labeled data, yet they often lack scalability. To bridge these gaps, this study introduces a novel MED-Prompt framework that employs pretrained models such as BERT, BioBERT, and ClinicalBERT. The core of our framework lies in developing specialized prompts, which act as guiding instructions for the models during the prediction process. MED-Prompt develops prompts that help models interpret and extract medical information from clinical corpus. The clinical text was derived from the widely known MIMIC-III 11https://physionet.org/content/mimiciii/1.4/. dataset. The study performs a comparative analysis and evaluates the performance of Manual-Prompt and GPT-Prompts. Further, a fine-tuned approach is developed within MED-Prompt, leveraging transfer learning to achieve prompt-guided medicine predictions. The proposed method achieved a maximum F1-score of 96.8%, which is more than 40% F1-score higher than the pretrained model. In addition, the fine-tuned also showed an average of 2.38 times better processing performance. These results revealed that MED-Prompt is scalable regarding the number of training records and input prompts. These results not only demonstrate the proficiency and effectiveness of the framework but also significantly reduce computational requirements. This also indicates that the proposed approach has the potential to significantly improve patient care, reduce resource requirements, and increase the overall effectiveness of AI-driven medical prediction systems.

Medical Information Extraction With NLP-Powered QABots: A Real-World Scenario.

The advent of computerized medical recording systems in healthcare facilities has made data retrieval tasks easier, compared to manual recording. Nevertheless, the potential of the information contained within medical records remains largely untapped, mostly due to the time and effort required to extract data from unstructured documents. Natural Language Processing (NLP) represents a promising solution to this challenge, as it enables the use of automated text-mining tools for clinical practitioners. In this work, we present the architecture of the Virtual Dementia Institute (IVD), a consortium of sixteen Italian hospitals, using the NLP Extraction and Management Tool (NEMT), a (semi-) automated end-to-end pipeline that extracts relevant information from clinical documents and stores it in a centralized REDCap database. After defining a common Case Report Form (CRF) across the IVD hospitals, we implemented NEMT, the core of which is a Question Answering Bot (QABot) based on a modern NLP model. This QABot is fine-tuned on thousands of examples from IVD centers. Detailed descriptions of the process to define a common minimum dataset, Inter-Annotator Agreement calculated on clinical documents, and NEMT results are provided. The best QABot performance show an Exact Match score (EM) of 78.1%, a F1-score of 84.7%, a Lenient Accuracy (LAcc) of 0.834, and a Mean Reciprocal Rank (MRR) of 0.810. EM and F1 scores outperform the same metrics obtained with ChatGPTv3.5 (68.9% and 52.5%, respectively). With NEMT the IVD has been able to populate a database that will contain data from thousands of Italian patients, all screened with the same procedure. NEMT represents an efficient tool that paves the way for medical information extraction and exploitation for new research studies.

Natural language processing for drug information extraction : Advancing knowledge discovery in biomedical literature

Natural language processing (NLP) has emerged as an important tool in the biomedical industry for bettering medication information extraction and other knowledge gathering. In this study, we explore the use of natural language processing (NLP) techniques to glean useful insights from massive volumes of biological literature, with the goal of better comprehending data pertaining to drugs. We want to automate the extraction of crucial aspects such drug interactions, side effects, and efficacy by utilizing sophisticated language models and semantic analysis. Effective and comprehensive drug information retrieval will be encouraged. Our innovation improves the drug development process by facilitating the rapid exploration of big databases for previously unknown connections and patterns. By facilitating the synthesis of data from several sources, NLP in biomedical research speeds up information extraction, which in turn accelerates drug development and improves patient care by allowing for evidence-based decision making. This research demonstrates the promising potential of natural language processing (NLP) for unraveling the complexities of drug-related data, ushering in a new era of learning in the biomedical field.

BIR: Biomedical Information Retrieval System for Cancer Treatment in Electronic Health Record Using Transformers

The rapid growth of electronic health records (EHRs) has led to unprecedented biomedical data. Clinician access to the latest patient information can improve the quality of healthcare. However, clinicians have difficulty finding information quickly and easily due to the sheer data mining volume. Biomedical information retrieval (BIR) systems can help clinicians find the information required by automatically searching EHRs and returning relevant results. However, traditional BIR systems cannot understand the complex relationships between EHR entities. Transformers are a new type of neural network that is very effective for natural language processing (NLP) tasks. As a result, transformers are well suited for tasks such as machine translation and text summarization. In this paper, we propose a new BIR system for EHRs that uses transformers for predicting cancer treatment from EHR. Our system can understand the complex relationships between the different entities in an EHR, which allows it to return more relevant results to clinicians. We evaluated our system on a dataset of EHRs and found that it outperformed state-of-the-art BIR systems on various tasks, including medical question answering and information extraction. Our results show that Transformers are a promising approach for BIR in EHRs, reaching an accuracy and an F1-score of 86.46%, and 0.8157, respectively. We believe that our system can help clinicians find the information they need more quickly and easily, leading to improved patient care.

Opportunities and challenges for ChatGPT and large language models in biomedicine and health.

ChatGPT has drawn considerable attention from both the general public and domain experts with its remarkable text generation capabilities. This has subsequently led to the emergence of diverse applications in the field of biomedicine and health. In this work, we examine the diverse applications of large language models (LLMs), such as ChatGPT, in biomedicine and health. Specifically, we explore the areas of biomedical information retrieval, question answering, medical text summarization, information extraction and medical education and investigate whether LLMs possess the transformative power to revolutionize these tasks or whether the distinct complexities of biomedical domain presents unique challenges. Following an extensive literature survey, we find that significant advances have been made in the field of text generation tasks, surpassing the previous state-of-the-art methods. For other applications, the advances have been modest. Overall, LLMs have not yet revolutionized biomedicine, but recent rapid progress indicates that such methods hold great potential to provide valuable means for accelerating discovery and improving health. We also find that the use of LLMs, like ChatGPT, in the fields of biomedicine and health entails various risks and challenges, including fabricated information in its generated responses, as well as legal and privacy concerns associated with sensitive patient data. We believe this survey can provide a comprehensive and timely overview to biomedical researchers and healthcare practitioners on the opportunities and challenges associated with using ChatGPT and other LLMs for transforming biomedicine and health.

Multi-aspect Understanding with Cooperative Graph Attention Networks for Medical Dialogue Information Extraction

Medical dialogue information extraction is an important but challenging task for Electronic Medical Records. Existing medical information extraction methods ignore the crucial information of sentence and multi-level dependency in dialogue, which limits their effectiveness for capturing essential medical information. To address these issues, we present a novel Multi-aspect Understanding with Cooperative Graph Attention Networks for Medical Dialogue Information Extraction to capture multi-aspect sentence information and multi-level dependency information from the dialogue. First, we propose the multi-aspect sentence encoder to capture various features from different perspectives. Second, we propose double graph attention networks to model the dependency features from intra-window and inter-window, respectively. Extensive experiments on a benchmark dataset have well-validated the effectiveness of the proposed method.

Machine Learning Applied to Cholesterol-Lowering Pharmacotherapy: Proof-of-Concept in High-Risk Patients Treated in Primary Care.

Objectives: Machine learning has potential to improve the management of lipid disorders. We explored the utility of machine learning in high-risk patients in primary care receiving cholesterol-lowering medications. Methods: Machine learning algorithms were created based on lipid management guidelines for England [National Institute for Health and Care Excellence (NICE) CG181] to reproduce the guidance with >95% accuracy. Natural language processing and therapy identification algorithms were applied to anonymized electronic records from six South London primary care general practices to extract medication information from free text fields. Results: Among a total of 48,226 adult patients, a subset of 5630 (mean ± standard deviation, age = 67 ± 13 years; male:female = 55:45) with a history of lipid-lowering therapy were identified. Additional major cardiometabolic comorbidities included type 2 diabetes in 13% (n = 724) and hypertension in 32% (n = 1791); all three risk factors were present in a further 28% (n = 1552). Of the 5630 patients, 4290 (76%) and 1349 (24%) were in primary and secondary cardiovascular disease prevention cohorts, respectively. Statin monotherapy was the most common current medication (82%, n = 4632). For patients receiving statin monotherapy, 71% (n = 3269) were on high-intensity therapy aligned with NICE guidance with rates being similar for the primary and secondary prevention cohorts. In the combined cohort, only 46% of patients who had been prescribed lipid-lowering therapy in the previous 12 months achieved the NICE treatment goal of >40% reduction in non-high-density lipoprotein cholesterol from baseline pretreatment levels. Based on the most recent data entry for patients not at goal the neural network recommended either increasing the dose of statin, adding complementary cholesterol-lowering medication, or obtaining an expert lipid opinion. Conclusions: Machine learning can be of value in (a) quantifying suboptimal lipid-lowering prescribing patterns, (b) identifying high-risk patients who could benefit from more intensive therapy, and (c) suggesting evidence-based therapeutic options.

Classification of Medical Image Notes for Image Labeling by Using MinBERT

The lack of labeled image data poses a serious challenge to the application of artificial intelligence (AI) in medical image diagnosis. Medical image notes contain valuable patient information that could be used to label images for machine learning tasks. However, most image note texts are unstructured with heterogeneity and short-paragraph characters, which fail traditional keyword-based techniques. We utilized a deep learning approach to recover missing labels for medical image notes automatically by using a combination of deep word embedding and deep neural network classifiers. Bidirectional encoder representations from transformers trained on medical image notes corpus (MinBERT) were proposed. We applied the proposed techniques to two typical classification tasks: Medical image type identification and clinical diagnosis identification. The two methods significantly outperformed baseline methods and presented high accuracies of 99.56% and 99.72% in image type identification and of 94.56% and 92.45% in clinical diagnosis identification. Visualization analysis further indicated that word embedding could efficiently capture semantic similarities and regularities across diverse expressions. Results indicated that our proposed framework could accurately recover the missing label information of medical images through the automatic extraction of electronic medical record information. Hence, it could serve as a powerful tool for exploring useful training data in various medical AI applications.

A unified framework of medical information annotation and extraction for Chinese clinical text

Medical information extraction consists of a group of natural language processing (NLP) tasks, which collaboratively convert clinical text to pre-defined structured formats. This is a critical step to exploit electronic medical records (EMRs). Given the recent thriving NLP technologies, model implementation and performance seem no longer an obstacle, whereas the bottleneck locates on a high-quality annotated corpus and the whole engineering workflow. This study presents an engineering framework consisting of three tasks, i.e., medical entity recognition, relation extraction and attribute extraction. Within this framework, the whole workflow is demonstrated from EMR data collection through model performance evaluation. Our annotation scheme is designed to be comprehensive and compatible between the multiple tasks. With the EMRs from a general hospital in Ningbo, China, and the manual annotation by experienced physicians, our corpus is of large scale and high quality. Built upon this Chinese clinical corpus, the medical information extraction system show performance that approaches human annotation. The annotation scheme, (a subset of) the annotated corpus, and the code are all publicly released, to facilitate further research.

Radiomics in gastrointestinal stromal tumours: an up-to-date review.

Gastrointestinal stromal tumours are rare mesenchymal neoplasms originating from the Cajal cells and represent the most common sarcomas in the gastroenteric tract. Symptoms may be absent or non-specific, ranging from fatigue and weight loss to acute abdomen. Nowadays endoscopy, echoendoscopy, contrast-enhanced computed tomography, magnetic resonance imaging and positron emission tomography are the main methods for diagnosis. Because of their rarity, these neoplasms may not be included immediately in the differential diagnosis of a solitary abdominal mass. Radiomics is an emerging technique that can extract medical imaging information, not visible to the human eye, transforming it into quantitative data. The purpose of this review is to demonstrate how radiomics can improve the already known imaging techniques by providing useful tools for the diagnosis, treatment, and prognosis of these tumours.

Contextualized medication information extraction using Transformer-based deep learning architectures

ObjectiveTo develop a natural language processing (NLP) system to extract medications and contextual information that help understand drug changes. This project is part of the 2022 n2c2 challenge. Materials and methodsWe developed NLP systems for medication mention extraction, event classification (indicating medication changes discussed or not), and context classification to classify medication changes context into 5 orthogonal dimensions related to drug changes. We explored 6 state-of-the-art pretrained transformer models for the three subtasks, including GatorTron, a large language model pretrained using > 90 billion words of text (including > 80 billion words from > 290 million clinical notes identified at the University of Florida Health). We evaluated our NLP systems using annotated data and evaluation scripts provided by the 2022 n2c2 organizers. ResultsOur GatorTron models achieved the best F1-scores of 0.9828 for medication extraction (ranked 3rd), 0.9379 for event classification (ranked 2nd), and the best micro-average accuracy of 0.9126 for context classification. GatorTron outperformed existing transformer models pretrained using smaller general English text and clinical text corpora, indicating the advantage of large language models. ConclusionThis study demonstrated the advantage of using large transformer models for contextual medication information extraction from clinical narratives.