Clinical Applications of Artificial Intelligence in COPD Care

This article was written by Kalli Mago.

Advances in artificial intelligence (AI) are changing many areas of life, including health care. New AI tools can improve efficiency and expand access to high-quality care. In the treatment of COPD, AI may support earlier diagnosis, predict exacerbations, and customize treatment plans and pulmonary rehabilitation programs. Together, these uses could improve health outcomes for millions of people worldwide.

Early Detection and Diagnosis

Early and accurate diagnosis is essential for ensuring patients receive evidence-based COPD care. Globally, COPD is one of the leading causes of death, yet it is often missed or misdiagnosed. A recent meta-analysis of more than two dozen studies found that COPD went undiagnosed in 14-26% of smokers with symptoms of lung disease. In addition, about 25% of patients using inhaled treatments for other conditions had COPD but had not been diagnosed.¹

AI may help address this diagnostic gap. Studies suggest that some AI models can diagnose COPD with accuracy rates between 80 and 100%.² While not yet mainstream, these tools show promise as added support in clinical practice.

The 2026 GOLD Report describes several AI-assisted approaches to diagnosing COPD that are currently under development.³ These include:

• Analysis of individual breath characteristics⁴
• Analysis of voice features⁵
• Oscillometry⁶
• Motion sensors to predict 6MWT⁷
• Chest computed tomography (CT)
• Electrical Impedance Tomography (EIT)⁸
• Analysis of data from wearable sensors, such as electronic nose signals

For example, AI can analyze recordings of patients saying certain vowel sounds or assess signals from motion sensors placed on the chest. By examining pronunciation, respiratory rate, and other key factors, AI can identify whether a person is likely to have COPD.⁹

Although many of these tools are still being studied, they represent promising options for the future. In particular, these new approaches may help patients in regions with limited access to specialized care.⁹

Exacerbation Prediction and Prevention

AI may also help identify COPD patients who are most at risk for exacerbations. This allows physicians to monitor patients more closely and take early steps to intervene.

One example is a machine learning model developed by Zeng et al. to predict severe COPD exacerbations within the next year. This AI model analyzed patients’ administrative and clinical data. Factors included patient age, sex, insurance status, medications, comorbidities, and more. The final model achieved an accuracy rate up to 90.33%.¹⁰

Researchers continue to study similar models and other AI-based methods to predict exacerbations. Some approaches involve using vital signs from wearable sensors, while others use patient-reported data.¹¹ By identifying flare-ups earlier, these AI tools can support timely treatment to reduce illness, hospitalizations, and death.⁹

Personalized Treatment Plans and Pulmonary Rehabilitation Programs

AI also creates opportunities to personalize COPD treatment. AI tools can analyze multiple data sources to help identify the best treatment for each patient. For example, AI may suggest a treatment plan based on:

• Clinical phenotypes and biomarkers
• Response to previous medications
• Lung function trends monitored at home
• CT-based imaging data
• Patient-reported symptoms⁷

One study found that using AI to help optimize patients’ treatment plans led to a 53% decrease in exacerbation risk.¹²

Similarly, integrating AI into pulmonary rehabilitation can allow for more personalized support and guidance. AI can help pulmonary rehab providers to deeply analyze patients’ movement and track adherence through wearable sensors.¹³

Initial findings showed a significant improvement in 6-minute walking distance after undergoing AI-assisted pulmonary rehab. This suggests that effectively integrating AI in this area may significantly improve COPD patients’ exercise capacity and pulmonary function.¹³

Risks and Limitations of AI

Despite its potential, AI in health care also comes with risks and limitations. The State of Clinical AI 2026 Report notes that more than 1,200 FDA-cleared tools and 350,000 consumer health apps have created a $70 billion market. However, most of these tools have not been thoroughly peer reviewed. About 95% of AI-based medical devices were cleared through the FDA’s 501(k) pathway, which relies on similarity to existing devices.¹⁴

Overall, generalizability is a concern with AI tools currently on the market. Many AI models are trained using single-center, small datasets. As a result, they may not properly represent rare diseases and specific populations, such as children.⁹ This limits how well these tools may perform in a real-world setting and raises concerns about bias and accuracy.

The State of Clinical AI Report also warns that AI tools may fail when faced with missing data, uncertainty, or changing conditions.¹⁴ In medicine, such failures can have serious consequences.

Because of these risks, clinicians should use AI carefully. Clinician expertise and judgment remain essential for safe, effective care.

When used responsibly, AI has the potential to enhance COPD care, from diagnosis to treatment and pulmonary rehabilitation. Staying informed about both the benefits and limitations of these tools will help clinicians prepare for a future where AI plays a larger, more integrated role in health care.