Systematic Literature Review of Artificial Intelligence-Based Models Predicting COPD Exacerbations
Author(s)
Tamas Agh, MSc, PhD, MD1, Job FM van Boven, PhD2, Przemyslaw Kardas, PhD, MD3, Agnes Nagy, MSc4, Judit Tittmann, MD4, Sándor Kovács, BA, MBA, MSc1, Ákos Bernard Józwiak, PhD1, Faten Amer, PhD4, Irene Mommers, MSc2, Bertalan Németh, PhD1, János G. Pitter, PhD1, Judit Józwiak-Hagymásy, MSc1.
1Center for HTA and Pharmacoeconomic Research, University of Pecs & Syreon Research Institute, Budapest, Hungary, 2University Medical Center Groningen, University of Groningen, Groningen, Netherlands, 3Medical University of Lodz, Lodz, Poland, 4Center for HTA and Pharmacoeconomic Research, University of Pecs, Pecs, Hungary.
1Center for HTA and Pharmacoeconomic Research, University of Pecs & Syreon Research Institute, Budapest, Hungary, 2University Medical Center Groningen, University of Groningen, Groningen, Netherlands, 3Medical University of Lodz, Lodz, Poland, 4Center for HTA and Pharmacoeconomic Research, University of Pecs, Pecs, Hungary.
OBJECTIVES: Chronic obstructive pulmonary disease (COPD) is a leading cause of morbidity and mortality worldwide, with acute exacerbations significantly worsening patient outcomes and increasing healthcare costs. This systematic literature review aimed to identify existing artificial intelligence (AI) based prediction models for COPD exacerbations, assess their methodological characteristics, predictive performance and clinical applicability.
METHODS: A systematic search was conducted in MEDLINE (via PubMed), Embase, and relevant grey literature sources. Title/abstract and full-text screening were performed independently by two reviewers. Data extraction covered population characteristics, data sources, AI model types, predictors used, validation methods, performance metrics and clinical implementation. Methodological quality of studies was assessed using the TRIPOD+AI checklist.
RESULTS: A total of 899 records were screened from which 44 studies were included in the data extraction after removing 210 duplicates and screening 689 titles and abstracts, followed by full-text assessment of 181 articles. The included studies varied widely in geographical origin, data sources, AI techniques, and validation methods. Only a minority of studies reported external validation and even less clinical implementation. None of the identified models incorporated medication non-adherence as a predictor, despite its well-established role in the risk of COPD exacerbations.
CONCLUSIONS: Despite growing interest in AI-based prediction models for COPD exacerbations, the field is marked by heterogeneity in model design and a lack of consistent validation standards. Most models demonstrated moderate predictive capabilities but were limited by insufficient external validation and poor generalizability. Future research should prioritize robust external validation, integration of medication adherence and behavioural data, and assessment of real-world implementation to enhance clinical utility of AI-based prediction models in COPD management.
METHODS: A systematic search was conducted in MEDLINE (via PubMed), Embase, and relevant grey literature sources. Title/abstract and full-text screening were performed independently by two reviewers. Data extraction covered population characteristics, data sources, AI model types, predictors used, validation methods, performance metrics and clinical implementation. Methodological quality of studies was assessed using the TRIPOD+AI checklist.
RESULTS: A total of 899 records were screened from which 44 studies were included in the data extraction after removing 210 duplicates and screening 689 titles and abstracts, followed by full-text assessment of 181 articles. The included studies varied widely in geographical origin, data sources, AI techniques, and validation methods. Only a minority of studies reported external validation and even less clinical implementation. None of the identified models incorporated medication non-adherence as a predictor, despite its well-established role in the risk of COPD exacerbations.
CONCLUSIONS: Despite growing interest in AI-based prediction models for COPD exacerbations, the field is marked by heterogeneity in model design and a lack of consistent validation standards. Most models demonstrated moderate predictive capabilities but were limited by insufficient external validation and poor generalizability. Future research should prioritize robust external validation, integration of medication adherence and behavioural data, and assessment of real-world implementation to enhance clinical utility of AI-based prediction models in COPD management.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
CO228
Topic
Clinical Outcomes, Health Service Delivery & Process of Care
Topic Subcategory
Clinical Outcomes Assessment
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Respiratory-Related Disorders (Allergy, Asthma, Smoking, Other Respiratory)