Exploring the Future of COPD Care With AI-Guided Therapy: Christopher Carlin, MB ChB, PhD

Captions were auto-generated.

Artificial intelligence (AI) and machine learning models may offer a powerful new way to identify high-risk patients with chronic obstructive pulmonary disease (COPD) who could benefit from biologic therapies, explained Christopher Carlin, MB ChB, PhD, BSc, a respiratory physician from Glasgow, Scotland. Today at the European Respiratory Society Congress 2025, Carlin detailed a project that leveraged large-scale clinical data to improve patient stratification and model treatment outcomes, “AI-Driven Identification of High-Risk COPD Patients for Biologic Therapy: Pathway Development Opportunities.”

As a respiratory physician, Carlin leads a clinical service for patients with COPD and severe respiratory failure. He is also clinical lead with West of Scotland Innovation Hub, where he runs a portfolio of projects focusing on data-assisted digital technologies; early, accurate diagnosis; and development and implementation of new treatment pathways for patients.

His team utilized routine clinical data from 2021 on over 30,000 patients with diagnosed COPD from west Scotland from 2021. Using these data, they developed, validated, and operationalized machine learning models to stratify patients by their risk profiles. The models incorporated specific data insights on laboratory results and prescribing patterns to pinpoint individuals who met the inclusion criteria for biologic therapy as defined in randomized clinical trials. This approach, Carlin explains, allows researchers to model the potential gains if trial outcomes were replicated in routine clinical practice.

Although the use of AI to guide real-time clinical therapy decisions is not yet standard practice, significant research is underway. Carlin mentioned the prospective clinical trial, DYNAMIC AI (NCT05914220), which presents live AI-generated risk scores to multidisciplinary teams to see if this information can lead to care optimization for patients. Similar efforts are being explored at a population level for patients who might be eligible for asthma biologics or other new targets.

Despite the potential of AI is this arena, Carlin emphasizes the need for caution, acknowledging concerns about the "black box" nature of AI—questions about what the AI is actually doing and whether it could amplify existing health inequalities. However, he noted that the data required for these models can also help identify and address such inequalities, potentially allowing for fairer approaches than current service pathways provide. Patient feedback on this work has been "overwhelmingly positive," with some expressing surprise that such data-driven methods were not already in use. Still, Carlin stressed that careful scrutiny and ongoing engagement with both patients and colleagues are essential as this technology continues to develop.