Medical Ontology Research Articles

Ontology extension by online clustering with large language model agents.

An ontology is a structured framework that categorizes entities, concepts, and relationships within a domain to facilitate shared understanding, and it is important in computational linguistics and knowledge representation. In this paper, we propose a novel framework to automatically extend an existing ontology from streaming data in a zero-shot manner. Specifically, the zero-shot ontology extension framework uses online and hierarchical clustering to integrate new knowledge into existing ontologies without substantial annotated data or domain-specific expertise. Focusing on the medical field, this approach leverages Large Language Models (LLMs) for two key tasks: Symptom Typing and Symptom Taxonomy among breast and bladder cancer survivors. Symptom Typing involves identifying and classifying medical symptoms from unstructured online patient forum data, while Symptom Taxonomy organizes and integrates these symptoms into an existing ontology. The combined use of online and hierarchical clustering enables real-time and structured categorization and integration of symptoms. The dual-phase model employs multiple LLMs to ensure accurate classification and seamless integration of new symptoms with minimal human oversight. The paper details the framework's development, experiments, quantitative analyses, and data visualizations, demonstrating its effectiveness in enhancing medical ontologies and advancing knowledge-based systems in healthcare.

Empowering healthcare education: A multilingual ontology for medical informatics and digital health (MIMO) integrated to artificial intelligence powered training in smart hospitals.

Objective: A comprehensive understanding of professional and technical terms is essential to achieving practical results in multidisciplinary projects dealing with health informatics and digital health. The medical informatics multilingual ontology (MIMO) initiative has been created through international cooperation. MIMO is continuously updated and comprises over 3700 concepts in 37 languages on the Health Terminology/Ontology Portal (HeTOP). Methods: We conducted case studies to assess the feasibility and impact of integrating MIMO into real-world healthcare projects. In HosmartAI, MIMO is used to index technological tools in a dedicated marketplace and improve partners' communication. Then, in SaNuRN, MIMO supports the development of a "Catalog and Index of Digital Health Teaching Resources" (CIDHR) backing digital health resources retrieval for health and allied health students. Results: In HosmartAI, MIMO facilitates the indexation of technological tools and smooths partners' interactions. In SaNuRN within CIDHR, MIMO ensures that students and practitioners access up-to-date, multilingual, and high-quality resources to enhance their learning endeavors. Conclusion: Integrating MIMO into training in smart hospital projects allows healthcare students and experts worldwide with different mother tongues and knowledge to tackle challenges facing the health informatics and digital health landscape to find innovative solutions improving initial and continuous education.

FuseLinker: Leveraging LLM’s pre-trained text embeddings and domain knowledge to enhance GNN-based link prediction on biomedical knowledge graphs

ObjectiveTo develop the FuseLinker, a novel link prediction framework for biomedical knowledge graphs (BKGs), which fully exploits the graph’s structural, textual and domain knowledge information. We evaluated the utility of FuseLinker in the graph-based drug repurposing task through detailed case studies. MethodsFuseLinker leverages fused pre-trained text embedding and domain knowledge embedding to enhance the graph neural network (GNN)-based link prediction model tailored for BKGs. This framework includes three parts: a) obtain text embeddings for BKGs using embedding-visible large language models (LLMs), b) learn the representations of medical ontology as domain knowledge information by employing the Poincaré graph embedding method, and c) fuse these embeddings and further learn the graph structure representations of BKGs by applying a GNN-based link prediction model. We evaluated FuseLinker against traditional knowledge graph embedding models and a conventional GNN-based link prediction model across four public BKG datasets. Additionally, we examined the impact of using different embedding-visible LLMs on FuseLinker’s performance. Finally, we investigated FuseLinker’s ability to generate medical hypotheses through two drug repurposing case studies for Sorafenib and Parkinson’s disease. ResultsBy comparing FuseLinker with baseline models on four BKGs, our method demonstrates superior performance. The Mean Reciprocal Rank (MRR) and Area Under receiver operating characteristic Curve (AUROC) for KEGG50k, Hetionet, SuppKG and ADInt are 0.969 and 0.987, 0.548 and 0.903, 0.739 and 0.928, and 0.831 and 0.890, respectively. ConclusionOur study demonstrates that FuseLinker is an effective novel link prediction framework that integrates multiple graph information and shows significant potential for practical applications in biomedical and clinical tasks. Source code and data are available at https://github.com/YKXia0/FuseLinker.

CMCN: Chinese medical concept normalization using continual learning and knowledge-enhanced

Medical Concept Normalization (MCN) is a crucial process for deep information extraction and natural language processing tasks, which plays a vital role in biomedical research. Although MCN in English has achieved significant research achievements, Chinese medical concept normalization (CMCN) remains insufficiently explored due to its complex syntactic structure and the paucity of Chinese medical semantic and ontology resources. In recent years, deep learning has been extensively applied across numerous natural language processing tasks, owing to its robust learning capabilities, adaptability, and transferability. It has proven to be well suited for intricate and specialized knowledge discovery research in the biomedical field. In this study, we conduct research on CMCN through the lens of deep learning. Specifically, our research introduces a model that leverages polymorphic semantic information and knowledge enhanced through multi-task learning and retain more important medical features through continual learning. As the cornerstone of CMCN, disease names are the main focus of this research. We evaluated various methodologies on Chinese disease dataset built by ourselves, finally achieving 76.12 % on Accuracy@1, 87.20 % on Accuracy@5 and 90.02 % on Accuracy@10 with our best-performing model GCBM-BSCL. This research not only advances the fields of knowledge mining and medical concept normalization but also enhances the integration and application of artificial intelligence in the medical and health field. We have published the source code and results on https://github.com/BearLiX/CMCN.

Enhancing Patient Empowerment and Health Literacy: Integrating Knowledge Graphs with Language Models for Personalized Health Content Delivery.

Health literacy empowers people to access, understand and apply health information to effectively manage their own health and to be an active participant in healthcare decisions. In this paper we propose a conceptual model for cognitive factors affecting health literacy and related socioeconomic aspects. Then we develop the HEALIE Knowledge Graph to represent the model, drawing from various medical ontologies, resources, and insights from domain experts. Finally, we combine the Knowledge Graph with a Large Language Model to generate personalised medical content and showcase the results through an example.

Leveraging Generative AI to Accelerate Biocuration of Medical Actions for Rare Disease.

Structured representations of clinical data can support computational analysis of individuals and cohorts, and ontologies representing disease entities and phenotypic abnormalities are now commonly used for translational research. The Medical Action Ontology (MAxO) provides a computational representation of treatments and other actions taken for the clinical management of patients. Currently, manual biocuration is used to assign MAxO terms to rare diseases, enabling clinical management of rare diseases to be described computationally for use in clinical decision support and mechanism discovery. However, it is challenging to scale manual curation to comprehensively capture information about medical actions for the more than 10,000 rare diseases. We present AutoMAxO, a semi-automated workflow that leverages Large Language Models (LLMs) to streamline MAxO biocuration for rare diseases. AutoMAxO first uses LLMs to retrieve candidate curations from abstracts of relevant publications. Next, the candidate curations are matched to ontology terms from MAxO, Human Phenotype Ontology (HPO), and MONDO disease ontology via a combination of LLMs and post-processing techniques. Finally, the matched terms are presented in a structured form to a human curator for approval. We used this approach to process 4,918 unique medical abstracts and identified annotations for 21 rare genetic diseases, we extracted 18,631 candidate disease-treatment curations, 538 of which were confirmed and transferred to the MAxO annotation dataset. The results of this project underscore the potential of generative AI to accelerate precision medicine by enabling a robust and comprehensive curation of the primary literature to represent information about diseases and procedures in a structured fashion. Although we focused on MAxO in this project, similar approaches could be taken for other biomedical curation tasks.

Predicting Future Disorders via Temporal Knowledge Graphs and Medical Ontologies.

Despite the vast potential for insights and value present in Electronic Health Records (EHRs), it is challenging to fully leverage all the available information, particularly that contained in the free-text data written by clinicians describing the health status of patients. The utilization of Named Entity Recognition and Linking tools allows not only for the structuring of information contained within free-text data, but also for the integration with medical ontologies, which may prove highly beneficial for the analysis of patient medical histories with the aim of forecasting future medical outcomes, such as the diagnosis of a new disorder. In this paper, we propose MedTKG, a Temporal Knowledge Graph (TKG) framework that incorporates both the dynamic information of patient clinical histories and the static information of medical ontologies. The TKG is used to model a medical history as a series of snapshots at different points in time, effectively capturing the dynamic nature of the patient's health status, while a static graph is used to model the hierarchies of concepts extracted from domain ontologies. The proposed method aims to predict future disorders by identifying missing objects in the quadruple 〈s, r, ?, t 〉, where s and r denote the patient and the disorder relation type, respectively, and t is the timestamp of the query. The method is evaluated on clinical notes extracted from MIMIC-III and demonstrates the effectiveness of the TKG framework in predicting future disorders and of medical ontologies in improving its performance.

Explanatory argumentation in natural language for correct and incorrect medical diagnoses

BackgroundA huge amount of research is carried out nowadays in Artificial Intelligence to propose automated ways to analyse medical data with the aim to support doctors in delivering medical diagnoses. However, a main issue of these approaches is the lack of transparency and interpretability of the achieved results, making it hard to employ such methods for educational purposes. It is therefore necessary to develop new frameworks to enhance explainability in these solutions.ResultsIn this paper, we present a novel full pipeline to generate automatically natural language explanations for medical diagnoses. The proposed solution starts from a clinical case description associated with a list of correct and incorrect diagnoses and, through the extraction of the relevant symptoms and findings, enriches the information contained in the description with verified medical knowledge from an ontology. Finally, the system returns a pattern-based explanation in natural language which elucidates why the correct (incorrect) diagnosis is the correct (incorrect) one. The main contribution of the paper is twofold: first, we propose two novel linguistic resources for the medical domain (i.e, a dataset of 314 clinical cases annotated with the medical entities from UMLS, and a database of biological boundaries for common findings), and second, a full Information Extraction pipeline to extract symptoms and findings from the clinical cases and match them with the terms in a medical ontology and to the biological boundaries. An extensive evaluation of the proposed approach shows the our method outperforms comparable approaches.ConclusionsOur goal is to offer AI-assisted educational support framework to form clinical residents to formulate sound and exhaustive explanations for their diagnoses to patients.

OntoMedRec: Logically-pretrained model-agnostic ontology encoders for medication recommendation

Recommending medications with electronic health records (EHRs) is a challenging task for data-driven clinical decision support systems. Most existing models learnt representations for medical concepts based on EHRs and make recommendations with the learnt representations. However, most medications appear in EHR datasets for limited times (the frequency distribution of medications follows power law distribution), resulting in insufficient learning of their representations of the medications. Medical ontologies are the hierarchical classification systems for medical terms where similar terms will be in the same class on a certain level. In this paper, we propose OntoMedRec, the logically-pretrained and model-agnostic medical Ontology Encoders for Medication Recommendation that addresses data sparsity problem with medical ontologies. We conduct comprehensive experiments on real-world EHR datasets to evaluate the effectiveness of OntoMedRec by integrating it into various existing downstream medication recommendation models. The result shows the integration of OntoMedRec improves the performance of various models in both the entire EHR datasets and the admissions with few-shot medications. We provide the GitHub repository for the source code. (https://github.com/WaicongTam/OntoMedRec)

Correlation Aware Relevance-Based Semantic Index for Clinical Big Data Repository.

In this paper, we focus on indexing mechanisms for unstructured clinical big integrated data repository systems. Clinical data is unstructured and heterogeneous, which comes in different files and formats. Accessing data efficiently and effectively are critical challenges. Traditional indexing mechanisms are difficult to apply on unstructured data, especially by identifying correlation information between clinical data elements. In this research work, we developed a correlation-aware relevance-based index that retrieves clinical data by fetching most relevant cases efficiently. In our previous work, we designed a methodology that categorizes medical data based on the semantics of data elements and merges them into an integrated repository. We developed a data integration system for medical data sources that combines heterogeneous medical data and provides access to knowledge-based database repositories to different users. In this research work, we designed an indexing system using semantic tags extracted from clinical data sources and medical ontologies that retrieves relevant data from database repositories and speeds up the process of data retrieval. Our objective is to provide an integrated biomedical database repository that can be used by radiologists as a reference, or for patient care, or by researchers. In this paper, we focus on designing a technique that performs data processing for data integration, learn the semantic properties of data elements, and develop a correlation-aware topic index that facilitates efficient data retrieval. We generated semantic tags by identifying key elements from integrated clinical cases using topic modeling techniques. We investigated a technique that identifies tags for merged categories and provides an index to fetch data from an integrated database repository. We developed a topic coherence matrix that shows how well a topic is supported by a corpus from clinical cases and medical ontologies. We were able to find more relevant results using an annotation index from an integrated database repository, and there was a 61% increase in a recall. We evaluated results with the help of experts and compared them with naive index (index with all terms from the corpus). Our approach improved data retrieval quality by providing most relevant results and reduced data retrieval time as we applied correlation-aware index on an integrated data repository. Topic indexing approach proposed in this research work identifies tags based on a correlation between different data elements, improves data retrieval time, and provides most relevant cases as an outcome of this system.

Creating a next-generation phenotype library: the health data research UK Phenotype Library.

To enable reproducible research at scale by creating a platform that enables health data users to find, access, curate, and re-use electronic health record phenotyping algorithms. We undertook a structured approach to identifying requirements for a phenotype algorithm platform by engaging with key stakeholders. User experience analysis was used to inform the design, which we implemented as a web application featuring a novel metadata standard for defining phenotyping algorithms, access via Application Programming Interface (API), support for computable data flows, and version control. The application has creation and editing functionality, enabling researchers to submit phenotypes directly. We created and launched the Phenotype Library in October 2021. The platform currently hosts 1049 phenotype definitions defined against 40 health data sources and >200K terms across 16 medical ontologies. We present several case studies demonstrating its utility for supporting and enabling research: the library hosts curated phenotype collections for the BREATHE respiratory health research hub and the Adolescent Mental Health Data Platform, and it is supporting the development of an informatics tool to generate clinical evidence for clinical guideline development groups. This platform makes an impact by being open to all health data users and accepting all appropriate content, as well as implementing key features that have not been widely available, including managing structured metadata, access via an API, and support for computable phenotypes. We have created the first openly available, programmatically accessible resource enabling the global health research community to store and manage phenotyping algorithms. Removing barriers to describing, sharing, and computing phenotypes will help unleash the potential benefit of health data for patients and the public.

KAMEL: Knowledge Aware Medical Entity Linkage to Automate Health Insurance Claims Processing

Automating the processing of health insurance claims to achieve "Straight-Through Processing" is one of the holy grails that all insurance companies aim to achieve. One of the major impediments to this automation is the difficulty in establishing the relationship between the underwriting exclusions that a policy has and the incoming claim's diagnosis information. Typically, policy underwriting exclusions are captured in free-text such as "Respiratory illnesses are excluded due to a pre-existing asthma condition". A medical claim coming from a hospital would have the diagnosis represented using the International Classification of Disease (ICD) codes from the World Health Organization. The complex and labour-intensive task of establishing the relationship between free-text underwriting exclusions in health insurance policies and medical diagnosis codes from health insurance claims is critical towards determining if a claim should be rejected due to underwriting exclusions. In this work, we present a novel framework that leverages both explicit and implicit domain knowledge present in medical ontologies and pre-trained language models respectively, to effectively establish the relationship between free-text describing medical conditions present in underwriting exclusions and the ICD-10CM diagnosis codes in health insurance claims. Termed KAMEL (Knowledge Aware Medical Entity Linkage), our proposed framework addresses the limitations faced by prior approaches when evaluated on real-world health insurance claims data. Our proposed framework have been deployed in several multi-national health insurance providers to automate their health insurance claims.

Abstract 5145: Association of a ctDNA biomarker of treatment response with clinical outcomes in a real-world pan-cancer cohort treated with tyrosine kinase inhibitors

Abstract Clinical evidence suggests that early changes in circulating tumor DNA tumor fraction (ctDNA TF) is predictive of patient response to immune checkpoint inhibitors, but the extent to which this approach can be used for patients on Tyrosine Kinase Inhibitors (TKIs) is not known. Here, we use xF Monitor to estimate ctDNA TF in patients that received TKIs and show that changes in circulating TF correlate with outcomes in a real-world pan-cancer cohort. Tempus xF Monitor is a ctDNA assay that measures quantitative molecular changes in ctDNA TF by utilizing diverse genomic events, dynamically weighting somatic variant allele frequencies and copy number variants, while using germline information to inform these estimates and providing single nucleotide variant (SNV) results on 105 genes. Molecular responders (MR) were defined as patients with ≥50% reduction in ctDNA TF between baseline and on-treatment time points, consistent with our previously established threshold. Deidentified patient records from the Tempus multimodal database were analyzed if patients had an xF test ≤ 15 weeks prior to the start of their first TKI regimen and an xF test 15 -180 days post-TKI initiation. Patients received TKI agents as classified in the Tempus Medical Ontology, which assigns a unique ID to key clinical concepts, validated and harmonized across industry standard terminologies. The TKI list was further validated using the NCI Metathesaurus (https://ncim.nci.nih.gov/ncimbrowser/) TKI concept (NCI Code C1967). Clinical endpoints were defined from TKI start to the first progression event or death (rwPFS) or death (rwOS), both censored on the last follow-up in event-free patients. Kaplan Meier analysis and Cox proportional hazard models were fitted to compare MR status and survival outcomes on TKI. The evaluable pan-cancer cohort N=69, median age = 61.5 yo, advanced/metastatic disease = 72%, 65% were female, consisted of > 10 cancer diagnoses (35% NSCLC, 26% BC), 41% 1L. In this cohort, 49% of patients received a TKI combination regimen, 21% with CT, 18% with ICI. 68% of patients with NGS testing received matched targeted mutation therapies based on NCCN guidelines. Patients classified as MR based on xF Monitor results had improved survival outcomes with a median rwOS of 951 days (vs 439 days for non-MRs). MR patients had significantly longer rwPFS (HR = 0.383, p = 0.005, CI = 0.12 - 0.75) and rwOS (HR = 0.205, p = 0.004, CI = 0.07 - 0.61) than non-MRs. xF Monitor is a novel tumor naive serial quantitative ctDNA TF algorithm off the Tempus xF assay that has the potential to be used clinically as a monitoring biomarker to stratify patients who are likely to benefit from TKI therapy. xF monitor can be used as a strategy to identify patients at a high risk of progression who could benefit from early switch or intensifying therapy. Results need to be prospectively validated in a larger cohort. Citation Format: Akash Mitra, John Guittar, Regina Schwind, Terri Driessen, Matthew Berginski, Christine Lo, Michelle M. Stein, Jonathan Freaney, Seung Won Hyun, Chithra Sangli, Kate Sasser, Halla Nimeiri, Aadel Chaudhuri. Association of a ctDNA biomarker of treatment response with clinical outcomes in a real-world pan-cancer cohort treated with tyrosine kinase inhibitors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5145.

Bridging semantics and medical ontology in the definition of bruxism: An argument for a holistic definition.

The objective of this commentary is to advocate for a holistic, ontology-based definition of bruxism. The intention is to synthesise the best aspects of current definitions into a structured ontological model, thereby refining and enhancing a comprehensive understanding of the full spectrum of bruxism. The commentary elaborates on the process of integrating these insights into a hierarchical ontology that aligns with ontological principles. Not directly applicable as this is a commentary. The proposed ontology-based definition of bruxism aims to clarify communication within the medical community and advance research by enabling a comprehensive ontology-based classification of bruxism. By aligning with ontological principles, this approach aspires to act as a catalyst for further research, discussion and consensus in the field.

Merging Ontologies and Data from Electronic Health Records

The Electronic Health Record (EHR) is a system for collecting and storing patient medical records as data that can be mechanically accessed, hence facilitating and assisting the medical decision-making process. EHRs exist in several formats, and each format lists thousands of keywords to classify patients data. The keywords are specific and are medical jargon; hence, data classification is very accurate. As the keywords constituting the formats of medical records express concepts by means of specific jargon without definitions or references, their proper use is left to clinicians and could be affected by their background, hence the interpretation of data could become slow or less accurate than that desired. This article presents an approach that accurately relates data in EHRs to ontologies in the medical realm. Thanks to ontologies, clinicians can be assisted when writing or analysing health records, e.g., our solution promptly suggests rigorous definitions for scientific terms, and automatically connects data spread over several parts of EHRs. The first step of our approach consists of converting selected data and keywords from several EHR formats into a format easier to parse, then the second step is merging the extracted data with specialised medical ontologies. Finally, enriched versions of the medical data are made available to professionals. The proposed approach was validated by taking samples of medical records and ontologies in the real world. The results have shown both versatility on handling data, precision of query results, and appropriate suggestions for relations among medical records.

Semi-automated ontology development scheme via text mining of scientific records

PurposeThe purpose of this study is to develop a domain independent, cost-effective, time-saving and semi-automated ontology generation framework that could extract taxonomic concepts from unstructured text corpus. In the human disease domain, ontologies are found to be extremely useful for managing the diversity of technical expressions in favour of information retrieval objectives. The boundaries of these domains are expanding so fast that it is essential to continuously develop new ontologies or upgrade available ones.Design/methodology/approachThis paper proposes a semi-automated approach that extracts entities/relations via text mining of scientific publications. Text mining-based ontology (TmbOnt)-named code is generated to assist a user in capturing, processing and establishing ontology elements. This code takes a pile of unstructured text files as input and projects them into high-valued entities or relations as output. As a semi-automated approach, a user supervises the process, filters meaningful predecessor/successor phrases and finalizes the demanded ontology-taxonomy. To verify the practical capabilities of the scheme, a case study was performed to drive glaucoma ontology-taxonomy. For this purpose, text files containing 10,000 records were collected from PubMed.FindingsThe proposed approach processed over 3.8 million tokenized terms of those records and yielded the resultant glaucoma ontology-taxonomy. Compared with two famous disease ontologies, TmbOnt-driven taxonomy demonstrated a 60%–100% coverage ratio against famous medical thesauruses and ontology taxonomies, such as Human Disease Ontology, Medical Subject Headings and National Cancer Institute Thesaurus, with an average of 70% additional terms recommended for ontology development.Originality/valueAccording to the literature, the proposed scheme demonstrated novel capability in expanding the ontology-taxonomy structure with a semi-automated text mining approach, aiming for future fully-automated approaches.

Translation of Ontological Concepts from English into German Using Commercial Translation Software and Expert Evaluation.

Medical ontologies are mostly available in English. This presents a language barrier that is a limitation in research and automated processing of patient data. The manual translation of ontologies is complex and time-consuming. However, there are commercial translation tools that have shown promising results in the field of medical terminology translation. The aim of this study is to translate selected terms of the Human Phenotype Ontology (HPO) from English into German using commercial translators. Six medical experts evaluated the translation candidates in an iterative process. The results show commercial translators, with DeepL in the lead, provide translations that are positively evaluated by experts. With a broader study scope and additional optimization techniques, commercial translators could support and facilitate the process of translating medical ontologies.

International Patient Summary Terminology.

SNOMED CT is a comprehensive medical ontology used in health care sectors across the world covering a wide range of concepts that support diversity at the point of healthcare. However, not all these concepts are needed for every use case; it is better to concentrate on those parts that apply to the particular application while preserving the meaning of relevant concepts. This paper considers the application of a novel subontology extraction method to create a new resource, called the IPS terminology, which functions as a standalone ontology with the same features as SNOMED CT, but is designed for cross-border patient care. The IPS terminology has been released for free use under an open license, with the intention of promoting interoperability of health information worldwide.

UMLS-Based Approach for Developing VoiS: Voice-Activated Conversational Agent for Self-Management of Multiple Chronic Conditions

This abstract proposes a system design for an ontology-based conversational agent (CA) for the self-management of chronic conditions. The proposed system plans to integrate the largest medical ontology, the Unified Medical Language System (UMLS) (Bodenreider, 2004), aiming to narrow the vocabulary gaps between health professionals and patients and to make the agent more responsive to the users.
 Recently, conversational agents (CAs) like ChatGPT, or computer dialog systems that simulate human-to-human communication in natural language, have risen and gained popularity in various health contexts (Bin Sawad et al., 2022; Montenegro et al., 2019) including self-management of chronic conditions (Griffin et al., 2020). Despite their potential, the currently available CAs are often criticized for lacking the capability to understand natural language inputs (Montenegro et al., 2019). This limitation can be highlighted in medical areas due to the known vocabulary gaps between health professionals and health consumers. The knowledge-grounded dialog flow for CAs presents the potential to lift this limitation, making CAs naturally converse with their users.
 In the proposed voice-activated self-monitoring support (VoiS) application, the research team plans to integrate the UMLS to make the agent better understand lay terms from patients and properly map those terms to medical concepts. This automated process is expected to improve the user experience in two folds: a) promote the quality of communication between the patients and health providers and b) make the VoiS app more responsive to user inputs, overcoming accepting only constrained user inputs (e.g., multiple choice of utterance).

Development and application of Chinese medical ontology for diabetes mellitus

ObjectiveTo develop a Chinese Diabetes Mellitus Ontology (CDMO) and explore methods for constructing high-quality Chinese biomedical ontologies.Materials and methodsWe used various data sources, including Chinese clinical practice guidelines, expert consensus, literature, and hospital information system database schema, to build the CDMO. We combined top-down and bottom-up strategies and integrated text mining and cross-lingual ontology mapping. The ontology was validated by clinical experts and ontology development tools, and its application was validated through clinical decision support and Chinese natural language medical question answering.ResultsThe current CDMO consists of 3,752 classes, 182 fine-grained object properties with hierarchical relationships, 108 annotation properties, and over 12,000 mappings to other well-known medical ontologies in English. Based on the CDMO and clinical practice guidelines, we developed 200 rules for diabetes diagnosis, treatment, diet, and medication recommendations using the Semantic Web Rule Language. By injecting ontology knowledge, CDMO enhances the performance of the T5 model on a real-world Chinese medical question answering dataset related to diabetes.ConclusionCDMO has fine-grained semantic relationships and extensive annotation information, providing a foundation for medical artificial intelligence applications in Chinese contexts, including the construction of medical knowledge graphs, clinical decision support systems, and automated medical question answering. Furthermore, the development process incorporated natural language processing and cross-lingual ontology mapping to improve the quality of the ontology and improved development efficiency. This workflow offers a methodological reference for the efficient development of other high-quality Chinese as well as non-English medical ontologies.