Search Results (28)

Background and Objectives: Dermatology relies on a complex terminology encompassing lesion types, distribution patterns, colors, and specialized sites such as hair and nails, while dermoscopy adds an additional descriptive framework, making interpretation subjective and challenging. Our study aims to evaluate the ability of a chatbot (Gemini 2) to generate dermatology descriptions across multiple languages and image types, and to assess the influence of prompt language on readability, completeness, and terminology consistency. Our research is based on the concept that non-English prompts are not mere translations of the English prompts but are independently generated texts that reflect medical and dermatological knowledge learned from non-English material used in the chatbot’s training. Materials and Methods: Five macroscopic and five dermoscopic images of common skin lesions were used. Images were uploaded to Gemini 2 with language-specific prompts requesting short paragraphs describing visible features and possible diagnoses. A total of 2400 outputs were analyzed for readability using LIX score and CLEAR (comprehensiveness, accuracy, evidence-based content, appropriateness, and relevance) assessment, while terminology consistency was evaluated via SNOMED CT mapping across English, French, German, and Greek outputs. Results: English and French descriptions were found to be harder to read and more sophisticated, while SNOMED CT mapping revealed the largest terminology mismatch in German and the smallest in French. English texts and macroscopic images achieved the highest accuracy, completeness, and readability based on CLEAR assessment, whereas dermoscopic images and non-English texts presented greater challenges. Conclusions: Overall, partial terminology inconsistencies and cross-lingual variations highlighted that the language of the prompt plays a critical role in shaping AI-generated dermatology descriptions. Full article

Background: Anatomopathological reports are largely unstructured, which limits automated data extraction, interoperability, and large-scale research. Manual extraction and standardization are costly and difficult to scale. Objective: We developed and evaluated an automated pipeline for entity extraction and multi-ontology normalization of anatomopathological reports. Methods: A corpus of 560 reports from the Military Hospital of Tunis, Tunisia, was manually annotated for three entity types: sample type, test performed, and finding. The entity extraction utilized BioBERT v1.1, while the normalization combined BioClinicalBERT multi-label classification with retrieval-augmented generation, incorporating both dense and BM25 sparse retrieval over SNOMED CT, LOINC, and ICD-11. The performance was measured using precision, recall, F1-score, and statistical tests. Results: BioBERT achieved high extraction performance (F1: 0.97 for the sample type, 0.98 for the test performed, and 0.93 for the finding; overall 0.963, 95% CI: 0.933–0.982), with low absolute errors. For terminology mapping, the combination of BioClinicalBERT and dense retrieval outperformed the standalone and BM25-based approaches (macro-F1: 0.6159 for SNOMED CT, 0.9294 for LOINC, and 0.7201 for ICD-11). Cohen’s Kappa ranged from 0.7829 to 0.9773, indicating substantial to near-perfect agreement. Conclusions: The pipeline provides robust automated extraction and multi-ontology coding of anatomopathological entities, supporting transformer-based named entity recognition with retrieval-augmented generation. However, given the limitations of this study, multi-institutional validation is needed before clinical deployment. Full article

Background/Purpose: Ophthalmic triage is challenging for non-specialists due to limited training and rising global eye disease burden. This study evaluates a multimodal framework integrating clinical text and ophthalmic imaging with large language models (LLMs). Textual consistency filtering and chain-of-thought (CoT) reasoning were incorporated to improve diagnostic accuracy. Methods: A dataset of 56 ophthalmology cases from a Singapore restructured hospital was pre-processed with acronym expansion, sentence reconstruction, and textual consistency filtering. To address dataset size limitations, 100 synthetic cases were generated via one-shot GPT-4 prompting, validated by semantic checks and ophthalmologist review. Three diagnostic approaches were tested: Text-Only, Image-Assisted, and Image with CoT. Diagnostic performance was quantified using a novel SNOMED-CT-based dissimilarity score, defined as the shortest path distance between predicted and reference diagnoses in the ontology, which was used to quantify semantic alignment. Results: The synthetic dataset included anterior segment (n = 40), posterior segment (n = 35), and extraocular (n = 25) cases. The text-only approach yielded a mean dissimilarity of 6.353 (95% CI: 4.668, 8.038). Incorporation of image assistance reduced this to 5.234 (95% CI: 3.930, 6.540), while CoT prompting provided further gains when imaging cues were ambiguous. Conclusions: The multimodal pipeline showed potential in improving diagnostic alignment in ophthalmology triage. Image inputs enhanced accuracy, and CoT reasoning reduced errors from ambiguous features, supporting its feasibility as a pilot framework for ophthalmology triage. Full article

The patient problem list is a key component of an electronic health record (EHR) and must be accurate and accessible for all professionals involved in patient care. Unfortunately, such a list is mostly found in an unstructured text format, is not easily sharable across digital health systems, and lacks semantic interoperability. Natural language processing (NLP) techniques are widely used for clinical concept extraction, particularly for English text. However, in the Canadian context, the clinical notes in a patient problem list can also be found in French. This research presents a framework based on Fast Healthcare Interoperability Resources (FHIR) consisting of an NLP clinical pipeline and a rule-based approach to converting the textual patient problem list, including notes regarding allergies, into an FHIR model. The proposed approach considers concept modifiers to map to the International Patient Summary (IPS) FHIR model element. The main contributions of this research include the early detection of FHIR resources from unstructured data written in the French language and the design of a rule-based algorithm to identify and map extracted data to the appropriate FHIR resource attributes using an annotator. A primary evaluation of the resource tag which uses the rule-based method demonstrates the feasibility of the proposed model to facilitate semantic interoperability. The assessment was conducted using the French FRASIMED corpora. Full article

The growing demand for accurate and efficient Chest X-Ray (CXR) interpretation has prompted the development of AI-driven systems to alleviate radiologist workload and reduce diagnostic variability. This paper introduces the Intelligent Humanized Radiology Analysis System (IHRAS), a modular framework that automates the end-to-end process of CXR analysis and report generation. IHRAS integrates four core components: (i) deep convolutional neural networks for multi-label classification of 14 thoracic conditions; (ii) Grad-CAM for spatial visualization of pathologies; (iii) SAR-Net for anatomical segmentation; and (iv) a large language model (DeepSeek-R1) guided by the CRISPE prompt engineering framework to generate structured diagnostic reports using SNOMED CT terminology. Evaluated on the NIH ChestX-ray dataset, IHRAS demonstrates consistent diagnostic performance across diverse demographic and clinical subgroups, and produces high-fidelity, clinically relevant radiological reports with strong faithfulness, relevancy, and alignment scores. The system offers a transparent and scalable solution to support radiological workflows while highlighting the importance of interpretability and standardization in clinical Artificial Intelligence applications. Full article

Background: Severe mental illness (SMI) is often linked to physical health issues, including multiple comorbidities. Physiotherapists are increasingly recognized for their role in addressing these health disparities. This study investigated the role of physiotherapy in managing physical health conditions in individuals admitted to inpatient mental health services. Objective: The primary aim was to examine the prevalence of physical comorbidities among individuals admitted to inpatient mental health services, comparing those referred to physiotherapy versus those not referred. Secondary aims included assessing the relationship between physiotherapy referral and admission duration and identifying patterns in referral likelihood based on primary and comorbid diagnoses. Methods: A data linkage analysis was conducted using records from Tees, Esk and Wear Valleys NHS Foundation Trust, encompassing admissions from September 2020 to January 2023. Demographic data, physiotherapy referral status, and SNOMED-CT coded diagnoses were analyzed. Results: Among 2150 admissions, 505 (23.5%) were referred for physiotherapy. Multimorbidity was present in 58.1% of admissions, with a higher prevalence (67.8%) in those referred to physiotherapy versus those not referred (55.1%). Individuals referred to physiotherapy had longer lengths of stay (117.3 days), compared to those not referred (44.1 days), suggesting that extended stays may indirectly facilitate the identification and management of physiotherapy needs. Referral likelihood was influenced by primary diagnoses and comorbidities. Conclusions: Approximately one in four inpatient admissions resulted in a physiotherapy referral, with a higher prevalence of multimorbidity in those referred. Further research is warranted to explore the long-term impacts of physiotherapy interventions on physical and mental health outcomes. Full article

Background/Objectives: We assessed the effectiveness and safety of vitamin K antagonists (VKAs) versus direct oral anticoagulants (DOACs) in patients with atrial fibrillation (AF) using artificial intelligence techniques. Methods: This is a retrospective study in 15 Spanish hospitals (2014–2020), including adult AF patients with no history of anticoagulation, thrombosis events, rheumatic mitral valvular heart disease, mitral valve stenosis, or pregnancy. We employed EHRead^® technology based on natural language processing (NLP) and machine learning (ML), along with SNOMED-CT terminology, to extract clinical data from electronic health records (EHRs). Using propensity score matching (PSM), the effectiveness, safety, and hospital mortality of VKAs versus DOACs were analyzed through Kaplan–Meier curves and Cox regression. Results: Out of 138,773,332 EHRs from 4.6 million individuals evaluated, 44,292 patients were included, 79.6% on VKAs and 20.4% on DOACs. Most patients were elderly [VKA 78 (70, 84) and DOAC 75 (66, 83) years], with numerous comorbidities (75.5% and 70.2% hypertension, 47.2% and 39.9% diabetes, and 40.3% and 34.8% heart failure, respectively). Additionally, 60.4% of VKA and 48.7% of DOAC users had a CHA2DS2-VASc Score ≥4. After PSM, 8929 patients per subgroup were selected. DOAC users showed a lower risk of thrombotic events [HR 0.81 (95% CI 0.70–0.94)], minor bleeding [HR 0.89 (95% CI 0.83–0.96)], and mortality [HR 0.80 (95% CI 0.69–0.92)]. Conclusions: Applying NLP and ML, we generated valuable real-world evidence on anticoagulated AF patients in Spain. Even in complex populations, DOACs have demonstrated a better safety and effectiveness profile than VKAs. Full article

We conducted a multicenter and retrospective study to describe the use of botulinum toxin type A (BoNT-A) to treat post-stroke spasticity (PSS). Data were extracted from free-text in electronic health records (EHRs) in five Spanish hospitals. We included adults diagnosed with PSS between January 2015 and December 2019, stratified into BoNT-A-treated and untreated groups. We used EHRead^® technology, which incorporates natural language processing and machine learning, as well as SNOMED CT terminology. We analyzed demographic data, stroke characteristics, BoNT-A use patterns, and other treatments. We reviewed the EHRs of 1,233,929 patients and identified 2190 people with PSS with a median age of 69 years; in total, 52.1% were men, 70.7% had cardiovascular risk factors, and 63.2% had suffered an ischemic stroke. Among the PSS patients, 25.5% received BoNT-A at least once. The median time from stroke to spasticity onset was 205 days, and the time from stroke to the first BoNT-A injection was 364 days. The primary goal of BoNT-A treatment was pain control. Among the study cohort, rehabilitation was the most common non-pharmacological treatment (95.5%). Only 3.3% had recorded monitoring scales. In conclusion, a quarter of patients with PSS received BoNT-A mainly for pain relief, typically one year after the stroke. Early treatment, disease monitoring, and better data documentation in EHRs are crucial to improve PSS patients’ care. Full article

This article presents an ingestion procedure towards an interoperable repository called ALPACS (Anonymized Local Picture Archiving and Communication System). ALPACS provides services to clinical and hospital users, who can access the repository data through an Artificial Intelligence (AI) application called PROXIMITY. This article shows the automated procedure for data ingestion from the medical imaging provider to the ALPACS repository. The data ingestion procedure was successfully applied by the data provider (Hospital Clínico de la Universidad de Chile, HCUCH) using a pseudo-anonymization algorithm at the source, thereby ensuring that the privacy of patients’ sensitive data is respected. Data transfer was carried out using international communication standards for health systems, which allows for replication of the procedure by other institutions that provide medical images. Objectives: This article aims to create a repository of 33,000 medical CT images and 33,000 diagnostic reports with international standards (HL7 HAPI FHIR, DICOM, SNOMED). This goal requires devising a data ingestion procedure that can be replicated by other provider institutions, guaranteeing data privacy by implementing a pseudo-anonymization algorithm at the source, and generating labels from annotations via NLP. Methodology: Our approach involves hybrid on-premise/cloud deployment of PACS and FHIR services, including transfer services for anonymized data to populate the repository through a structured ingestion procedure. We used NLP over the diagnostic reports to generate annotations, which were then used to train ML algorithms for content-based similar exam recovery. Outcomes: We successfully implemented ALPACS and PROXIMITY 2.0, ingesting almost 19,000 thorax CT exams to date along with their corresponding reports. Full article

Background: Mental health (MH) polypharmacy, defined as prescribing multiple mental health medications for the same condition, presents significant challenges in clinical practice. With varying prevalence rates and an increasing trend, particularly in the UK, this deprescribing prospective quality improvement project aimed to address the complexities and risks associated with MH polypharmacy. Patients and Methods: A large primary care centre in London was selected for this project. Electronic records of 667 patients (non-coded in mental health lists) were analysed as a result of the absence of a Systematised Nomenclature of Medicine Clinical Terms (SNOMED CT) for mental health. Seventy-two non-coded patients exhibiting “same-class” as well as “adjunctive” and “augmentation” polypharmacy were identified. Their demographic and health data, including MH diagnoses, physical status, and lifestyle habits, were evaluated. This deprescribing prospective project included 68 patients and employed a model inspired by the Plan–Do–Study–Act (PDSA) cycle, focusing on reducing psychotropic, adjunctive, and augmentative medications while monitoring mental health control through face-to-face consultations using the Patient Health Questionnaire-9 (PHQ-9) and Generalised Anxiety Disorder Assessment-7 (GAD-7) scores, alongside physical health parameters. Results: The project revealed a significant decrease in the average number of psychotropic and adjunct medications from initial consultations to the end of the 18-month period. Additionally, a marked reduction in reported side effects and drug interactions was observed. Improvements in mental health control, as evidenced by PHQ-9 and GAD-7 scores, were noted. Physical health parameters, including BMI, blood pressure, heart rate, HbA1c, and cholesterol levels, also showed significant improvements. Educational initiatives for patients and clinicians were successfully implemented, contributing to these positive outcomes. Discussion: The project faced challenges like balancing medication reduction with mental health stability, patient apprehension, and the absence of standardised protocols. However, the successful reduction in medication numbers and the improvement in health outcomes highlight the effectiveness of the model. This project underscores the necessity of a tailored approach to MH polypharmacy, emphasising continuous education, clinical titration, and adherence to guidelines. Future research is needed to develop clear guidelines for medication combination in mental health care and to understand the long-term effects of polypharmacy in mental health populations. Conclusions: This project demonstrates the potential for significant improvements in the management of MH polypharmacy. By carefully managing medication reductions and employing a comprehensive care approach, including patient education and clinician training, the project achieved improvements in both mental and physical health outcomes. These findings suggest a promising direction for future practices in MH polypharmacy management. Full article

The cross-institutional secondary use of medical data benefits from structured semantic annotation, which ideally enables the matching and merging of semantically related data items from different sources and sites. While numerous medical terminologies and ontologies, as well as some tooling, exist to support such annotation, cross-institutional data usage based on independently annotated datasets is challenging for multiple reasons: the annotation process is resource intensive and requires a combination of medical and technical expertise since it often requires judgment calls to resolve ambiguities resulting from the non-uniqueness of potential mappings to various levels of ontological hierarchies and relational and representational systems. The divergent resolution of such ambiguities can inhibit joint cross-institutional data usage based on semantic annotation since data items with related content from different sites will not be identifiable based on their respective annotations if different choices were made without further steps such as ontological inference, which is still an active area of research. We hypothesize that a collaborative approach to the semantic annotation of medical data can contribute to more resource-efficient and high-quality annotation by utilizing prior annotational choices of others to inform the annotation process, thus both speeding up the annotation itself and fostering a consensus approach to resolving annotational ambiguities by enabling annotators to discover and follow pre-existing annotational choices. Therefore, we performed a requirements analysis for such a collaborative approach, defined an annotation workflow based on the requirement analysis results, and implemented this workflow in a prototypical Collaborative Annotation Tool (CoAT). We then evaluated its usability and present first inter-institutional experiences with this novel approach to promote practically relevant interoperability driven by use of standardized ontologies. In both single-site usability evaluation and the first inter-institutional application, the CoAT showed potential to improve both annotation efficiency and quality by seamlessly integrating collaboratively generated annotation information into the annotation workflow, warranting further development and evaluation of the proposed innovative approach. Full article

Semantic web technologies (SWT) represent data in a format that is easier for machines to understand. Validating the knowledge in data graphs created using SWT is critical to ensure that the axioms accurately represent the so-called “real” world. However, data graph validation is a significant challenge in the semantic web domain. The Shapes Constraint Language (SHACL) is the latest W3C standard developed with the goal of validating data-graphs. SHACL (pronounced as shackle) is a relatively new standard and hitherto has predominantly been employed to validate generic data graphs like WikiData and DBPedia. In generic data graphs, the name of a class does not affect the shape of a class, but this is not the case with biomedical ontology data graphs. The shapes of classes in biomedical ontology data graphs are highly influenced by the names of the classes, and the SHACL shape creation methods developed for generic data graphs fail to consider this characteristic difference. Thus, the existing SHACL shape creation methods do not perform well for domain-specific biomedical ontology data graphs. Maintaining the quality of biomedical ontology data graphs is crucial to ensure accurate analysis in safety-critical applications like Electronic Health Record (EHR) systems referencing such data graphs. Thus, in this work, we present a novel method to create enhanced SHACL shapes that consider the aforementioned characteristic difference to better validate biomedical ontology data graphs. We leverage the knowledge available from lexical auditing techniques for biomedical ontologies and incorporate this knowledge to create smart SHACL shapes. We also create SHACL shapes (baseline SHACL graph) without incorporating the lexical knowledge of the class names, as is performed by existing methods, and compare the performance of our enhanced SHACL shapes with the baseline SHACL shapes. The results demonstrate that the enhanced SHACL shapes augmented with lexical knowledge of the class names identified 176 violations which the baseline SHACL shapes, void of this lexical knowledge, failed to detect. Thus, the enhanced SHACL shapes presented in this work significantly improve the validation performance of biomedical ontology data graphs, thereby reducing the errors present in such data graphs and ensuring safe use in the life-critical applications referencing them. Full article

This paper examines the status of the development of national digital health information systems (HIS) in Commonwealth of Independent States (CIS) member states. Data for research were collected using a questionnaire adapted from the questionnaire of the WHO’s Third Global Survey on eHealth. The results showed that the digital transformation of HIS was occurring in all seven CIS member states (participating in the study), which were financed by different resources. Laws and regulations on electronic medical records (EMR) use were present in almost all participating CIS member states. Various international standards and classifications were used to support development and the interoperability of digital health information system (d-HIS), including International Classification of Diseases (ICD), Digital Imaging and Communications in Medicine (DICOM), ISO 18308, Logical Observation Identifiers, Names, and Codes (LOINC), Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT), and ISO TC 215. Several CIS member states had adopted a national information security strategy for the safe processing of both personal data and medical confidential information. The digital transformation of healthcare and the Empowerment through Digital Health initiative are taking place in all CIS member states, which are at different stages of introducing electronic medical and health records. Full article

The standardization of data from medical studies and hospital information systems to a common data model such as the Observational Medical Outcomes Partnership (OMOP) model can help make large datasets available for analysis using artificial intelligence approaches. Commonly, automatic mapping without intervention from domain experts delivers poor results. Further challenges arise from the need for translation of non-English medical data. Here, we report the establishment of a mapping approach which automatically translates German data variable names into English and suggests OMOP concepts. The approach was set up using study data from the Hamburg City Health Study. It was evaluated against the current standard, refined, and tested on a separate dataset. Furthermore, different types of graphical user interfaces for the selection of suggested OMOP concepts were created and assessed. Compared to the current standard our approach performs slightly better. Its main advantage lies in the automatic processing of German phrases into English OMOP concept suggestions, operating without the need for human intervention. Challenges still lie in the adequate translation of nonstandard expressions, as well as in the resolution of abbreviations into long names. Full article

Expressive clinical terminologies are of utmost importance for achieving a semantically interoperable data exchange and reuse in healthcare. SNOMED CT, widely respected as the most comprehensive terminology in medicine, provides formal concept definitions based on description logic which not only allows for advanced querying of SNOMED-CT-coded data but also for flexibly augmenting its 350,000 concepts by allowing a controlled combination of these. This ability for postcoordination largely increases the expressivity of the terminology but correlates with an intrinsic complexity. Complicated by the current lack of tooling support, postcoordination is widely either ignored or applied in an error-prone way. To help facilitate the adoption of postcoordination, we implemented a web application that guides users through the creation of postcoordinated expressions (PCEs) while ensuring adherence to syntactic and semantic constraints. Our approach was largely facilitated by making use of the extensive SNOMED CT specifications as well as advanced HL7 FHIR Terminology Services. Qualitative evaluations confirmed the usability of the developed application and the correctness of the PCEs created with it. Full article