Pub Date : 2026-01-01DOI: 10.1016/j.rceng.2025.502413
J. Mora-Delgado , L. Ramos-Ruperto , M.J. Pardilla , M.Á. Sicilia , A. Rodríguez-González , J.M Sempere , R. Puchades
This work aims to provide internists with a practical, focused overview of how generative AI based on large language models can be effectively integrated into daily clinical practice. It describes the primary adaptation mechanisms like fine-tuning and retrieval-augmented generation (RAG) for tasks such as report generation, synthesis of clinical findings, and support in differential diagnoses, highlighting real-world examples in Internal Medicine. Technical and organizational requirements for adoption are analyzed, including computing infrastructure, integration with electronic health records, and security/privacy protocols under GDPR and the EU AI Act. Opportunities for enhancing clinical decision-making, optimizing workflows, and reducing administrative burden are emphasized, alongside current limitations like bias, hallucinations, and the need for human oversight. Finally, recommendations are offered for prospective validation in real-world settings and for ensuring explainable transparency, with the goal of empowering internists to incorporate these innovative tools responsibly and efficiently.
{"title":"Generative AI: foundational models. Natural Language Processing (NLP) and LARGE Language Models (LLM)","authors":"J. Mora-Delgado , L. Ramos-Ruperto , M.J. Pardilla , M.Á. Sicilia , A. Rodríguez-González , J.M Sempere , R. Puchades","doi":"10.1016/j.rceng.2025.502413","DOIUrl":"10.1016/j.rceng.2025.502413","url":null,"abstract":"<div><div>This work aims to provide internists with a practical, focused overview of how generative AI based on large language models can be effectively integrated into daily clinical practice. It describes the primary adaptation mechanisms like fine-tuning and retrieval-augmented generation (RAG) for tasks such as report generation, synthesis of clinical findings, and support in differential diagnoses, highlighting real-world examples in Internal Medicine. Technical and organizational requirements for adoption are analyzed, including computing infrastructure, integration with electronic health records, and security/privacy protocols under GDPR and the EU AI Act. Opportunities for enhancing clinical decision-making, optimizing workflows, and reducing administrative burden are emphasized, alongside current limitations like bias, hallucinations, and the need for human oversight. Finally, recommendations are offered for prospective validation in real-world settings and for ensuring explainable transparency, with the goal of empowering internists to incorporate these innovative tools responsibly and efficiently.</div></div>","PeriodicalId":94354,"journal":{"name":"Revista clinica espanola","volume":"226 1","pages":"Article 502413"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145812653","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.rceng.2025.502421
J. García Alegría , C. García Tortosa , M.D. Martín Escalante , F. Miralles Linares , R. Puchades-Rincón de Arellano , M.M. Chimeno-Viñas
Background/objective
Artificial intelligence (AI) has been revolutionising medical practice in recent years. The aim of this study was to analyze the perception of self-knowledge, personal experience, degree of use and training needs in AI among Spanish internists who are members of the Spanish Society of Internal Medicine (SEMI) in order to guide their educational activities.
Materials and methods
Cross-sectional study using an anonymous survey with demographic variables, categorical questions, multiple-choice questions, and open-ended qualitative questions. Descriptive analysis with differences between age groups. The minimum estimated sample size of representative members was 368.
Results
504 valid responses were analyzed (82% specialists, 16% residents). Self-perceived knowledge of AI was mainly intermediate or basic, with higher levels among those under 30 and lower levels among those over 60. Three out of four respondents had used AI, mainly in clinical practice, followed by research and teaching. The main perceived barriers were lack of specific training, doubts about reliability and ethical-legal issues, as well as technological limitations and resistance to change. The vast majority considered AI training to be important or very important, with particular interest in practical clinical applications, basic fundamentals and tool evaluation. The willingness to incorporate AI into practice was high across all age groups.
Conclusions
Spanish internists have varying levels of knowledge about artificial intelligence, with younger doctors having greater knowledge, and its main current use is in clinical practice. Lack of training is the main barrier to its incorporation, despite high demand for training and a general willingness to adopt it, highlighting the need for training programs and strategies for integrating AI into internal medicine.
{"title":"Artificial intelligence in internal medicine: knowledge, clinical use and training needs","authors":"J. García Alegría , C. García Tortosa , M.D. Martín Escalante , F. Miralles Linares , R. Puchades-Rincón de Arellano , M.M. Chimeno-Viñas","doi":"10.1016/j.rceng.2025.502421","DOIUrl":"10.1016/j.rceng.2025.502421","url":null,"abstract":"<div><h3>Background/objective</h3><div>Artificial intelligence (AI) has been revolutionising medical practice in recent years. The aim of this study was to analyze the perception of self-knowledge, personal experience, degree of use and training needs in AI among Spanish internists who are members of the Spanish Society of Internal Medicine (SEMI) in order to guide their educational activities.</div></div><div><h3>Materials and methods</h3><div>Cross-sectional study using an anonymous survey with demographic variables, categorical questions, multiple-choice questions, and open-ended qualitative questions. Descriptive analysis with differences between age groups. The minimum estimated sample size of representative members was 368.</div></div><div><h3>Results</h3><div>504 valid responses were analyzed (82% specialists, 16% residents). Self-perceived knowledge of AI was mainly intermediate or basic, with higher levels among those under 30 and lower levels among those over 60. Three out of four respondents had used AI, mainly in clinical practice, followed by research and teaching. The main perceived barriers were lack of specific training, doubts about reliability and ethical-legal issues, as well as technological limitations and resistance to change. The vast majority considered AI training to be important or very important, with particular interest in practical clinical applications, basic fundamentals and tool evaluation. The willingness to incorporate AI into practice was high across all age groups.</div></div><div><h3>Conclusions</h3><div>Spanish internists have varying levels of knowledge about artificial intelligence, with younger doctors having greater knowledge, and its main current use is in clinical practice. Lack of training is the main barrier to its incorporation, despite high demand for training and a general willingness to adopt it, highlighting the need for training programs and strategies for integrating AI into internal medicine.</div></div>","PeriodicalId":94354,"journal":{"name":"Revista clinica espanola","volume":"226 1","pages":"Article 502421"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145829533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.rceng.2025.502418
A. Carrasco Laraña , J. Álvarez Troncoso , J.J. Ríos Blanco
Introduction
Systemic autoimmune diseases (SADs) pose a diagnostic challenge due to the heterogeneity of their manifestations and the frequent overlap of symptoms. The integration of large language models (LLMs), such as GPT-4, could complement clinical judgment through the systematic analysis of standardized clinical data.
Objective
To evaluate the diagnostic capability of GPT-4 in patients with SADs at a tertiary care center, comparing its results with the final consensus diagnosis issued by specialists.
Methods
A retrospective study was conducted on a cohort of 101 consecutively treated patients between January 1 and March 31, 2024, at the SAD Unit of La Paz University Hospital. Data collection was carried out using the unit's standardized anamnesis protocol. The “my GPT” model, based on GPT-4 and trained according to international diagnostic criteria, was evaluated following TRIPOD‐AI guidelines.
Results
The overall diagnostic accuracy rate was 97.03%. Analysis based solely on anamnesis data achieved an accuracy of 82.18%, which increased by 14.85% when immunological results were included. A 100% accuracy was achieved in diagnosing systemic lupus erythematosus, Sjögren's syndrome, inflammatory myopathies, Behçet's disease, and scleroderma. In contrast, for sarcoidosis and vasculitis—conditions that often require histological confirmation—accuracy was 91.67% and 80%, respectively.
Conclusion
The use of GPT-4, grounded in systematic clinical data collection and evaluated in accordance with TRIPOD‐AI guidelines, demonstrates strong potential as an auxiliary tool in the diagnosis of SADs. Integrating this approach into clinical practice could help reduce interobserver variability and optimize decision-making.
{"title":"Integration of natural language models in the diagnosis of systemic autoimmune diseases: validation of GPT-4 in a tertiary care center","authors":"A. Carrasco Laraña , J. Álvarez Troncoso , J.J. Ríos Blanco","doi":"10.1016/j.rceng.2025.502418","DOIUrl":"10.1016/j.rceng.2025.502418","url":null,"abstract":"<div><h3>Introduction</h3><div>Systemic autoimmune diseases (SADs) pose a diagnostic challenge due to the heterogeneity of their manifestations and the frequent overlap of symptoms. The integration of large language models (LLMs), such as GPT-4, could complement clinical judgment through the systematic analysis of standardized clinical data.</div></div><div><h3>Objective</h3><div>To evaluate the diagnostic capability of GPT-4 in patients with SADs at a tertiary care center, comparing its results with the final consensus diagnosis issued by specialists.</div></div><div><h3>Methods</h3><div>A retrospective study was conducted on a cohort of 101 consecutively treated patients between January 1 and March 31, 2024, at the SAD Unit of La Paz University Hospital. Data collection was carried out using the unit's standardized anamnesis protocol. The “my GPT” model, based on GPT-4 and trained according to international diagnostic criteria, was evaluated following TRIPOD‐AI guidelines.</div></div><div><h3>Results</h3><div>The overall diagnostic accuracy rate was 97.03%. Analysis based solely on anamnesis data achieved an accuracy of 82.18%, which increased by 14.85% when immunological results were included. A 100% accuracy was achieved in diagnosing systemic lupus erythematosus, Sjögren's syndrome, inflammatory myopathies, Behçet's disease, and scleroderma. In contrast, for sarcoidosis and vasculitis—conditions that often require histological confirmation—accuracy was 91.67% and 80%, respectively.</div></div><div><h3>Conclusion</h3><div>The use of GPT-4, grounded in systematic clinical data collection and evaluated in accordance with TRIPOD‐AI guidelines, demonstrates strong potential as an auxiliary tool in the diagnosis of SADs. Integrating this approach into clinical practice could help reduce interobserver variability and optimize decision-making.</div></div>","PeriodicalId":94354,"journal":{"name":"Revista clinica espanola","volume":"226 1","pages":"Article 502418"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145812649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号