Summary
COPD (Chronic obstructive pulmonary disease ) is a progressive respiratory disease associated with a high prevalence of psychiatric comorbidities (Moretta et al., 2024). Coexisting mental illnesses, for instance depression and anxiety, are common comorbidities in COPD patients. Anxiety and/or depression in COPD patients is linked with higher exacerbation rates, increase length of hospital stay, decrease in quality of life as well as functional status and increase mortality. There is still no consensus on the best methodological approach in the screening and treatment of anxiety and depression in this group of patients (Martínez-Gestoso et al., 2022). Contemporary advancement in Artificial Intelligence (AI) and Machine Learning (ML) are promising tools to predict mental-health risks among COPD patients (Pozza et al., 2025).
This blog describes the epidemiological impact of psychiatric comorbidities in COPD, explains AI-based predictive models in the real world and identifies the avenues to clinical integration.
Introduction: What is COPD?
COPD is a disease state recognized by irreversible airflow limitation caused by exposure to noxious agents, such as tobacco smoke and air pollution (GOLD Criteria). In developing countries exposure to biomass fuel contributes to COPD in females. COPD is characterized by shortness of breath, cough and wheezing; however, it can also cause significant systemic effects (Ostroff & BCACP, 2021). Around the globe, COPD affects over 210 million people and is a preeminent cause of morbidity and mortality, which increases both social and economic burden. COPD-associated mortality has doubled in the last 30 years (Ayilya & Nazeer, 2023).
Burden of Anxiety and depression in COPD
COPD patients are two to three times more likely to develop mental health problems as compared to the general population. Literature found anxiety and depression with COPD are 40% and 36% prevalent, respectively depending on disease stage and setting (Siraj, 2025).
Anxiety may manifest as acute panic during exacerbation or as chronic worry about disease progression. It is a significant prognostic indicator of hospitalization and readmission for acute exacerbation of COPD. Depression is a mood disorder that leads to long-term sadness as well as loss of interest in valued activities (Martínez-Gestoso et al., 2022). Common features of depression include feeling of emptiness, sadness, and irritable mood, followed by cognitive and somatic changes that considerably affect a person’s ability to function.
These psychiatric comorbidities are associated with lack of self-efficacy or self-confidence; as a result, poor self-care ability and poor coping skills related to the disease. These indicate poor adherence to COPD treatment, decreased health-related quality of life, increased functional impairment, increased risk of acute exacerbation of COPD, and death. Although mental health screening guidelines exist in the management of COPD, it is unevenly implemented in practice (ADELOWO, 2023).
Why AI Based Predictive Models for Anxiety and depression in COPD?
Conventional screening tools, including PHQ-9 and GAD-7, are underutilized because of time-consuming, self-reporting, and clinician workload. Moreover, they may overlook patients who are at risk but not symptomatic yet. Thus, Artificial intelligence and Machine Learning models offer a paradigm shift because:
- They utilize immense epidemiological and clinical data (EHRs, surveys, telehealth inputs).
- They can anticipate risk prior to the manifestation of symptoms based on trends in demographics, comorbidities, and behaviors.
- They enhance efficiency, scalability and personalization of COPD care.
AI Based Predictive Models-Real World Scenarios/Evidence based Studies
| Model | Data sources | Predictors | Methods | Results | Implementation |
| SVM-Based Depression Risk Model
(Feng et al., 2025) |
Examined 1,638 COPD patients (NHANES database (2005-2018). | sleep disturbance, age, poverty ratio, hypertension, cardiovascular disease, and other comorbidities | Nine ML models were compared. Performance was assessed with AUC analysis | The SVM model performed best (AUC ~0.89). SHAP analysis showed poor sleep quality, younger age, and low socioeconomic status as leading risk drivers | web-based risk calculator was developed which enable rapid depression risk assessment in COPD patients |
| XGBoost Depression risk Predictive Model
(Zhao et al., 2025) |
Examined 2,921 patients with COPD (CHARLS data) | 11 predictors were chosen that comprised of socio-demographics, lifestyle, and comorbidities | Compared to 6 algorithms (SVM, MLP, RF, LightGBM, XGBoost) | XGBoost got the AUROC 0.811 (training) and 0.748 (temporal validation) | Developed web-based prediction platform for clinical implementation |
How does AI based Models Work?
COPD related depression AI models are trained on large scale data (e.g., NHANES, CHARLS), which contain clinical, demographic, lifestyle, and behavioral data. The most influential predictors, including sleep quality, comorbidities, socioeconomic status, and lifestyle factors are determined through feature selection methods (0-1) such as LASSO and Boruta. Machine learning algorithms (e.g. SVM, XGBoost, logistic regression, or voice-based acoustic models) process these inputs to approximate individual depression risk. Interpretability may be enhanced with tools such as SHAP or nomograms, whereas web calculators ensure results are available to clinical use (Feng et al., 2025; Zhao et al., 2025).
Clinical Integration
Practically, the integration of AI based models into COPD care can be done in a stepwise manner. Initially, patients with high-risk scores calculated by AI algorithms are subjected to confirmatory screening using some proven tools like PHQ 9 or GAD 7. Depending on severity, management may then be customized: such as, mild cases can be treated with digital cognitive behavioral therapy (CBT), pulmonary rehabilitation or structured counseling, whereas moderate to severe cases undergo a referral to psychiatric testing and, when necessary, pharmacological therapy. Another use of AI is remote monitoring. Lastly, the decision support systems provide clinicians with risk scores and SHAP based explanations (e.g., high risk due to sleep disorder, frequent exacerbations, and low income), so that predictions can be transparent and executable in clinical workflows.
Ethical Consideration
Data privacy, bias as well as equitable access are crucial because underrepresented populations would face inaccurate predictions. Furthermore, integration in clinical setting should balance AI driven recommendation with human judgment to avoid overreliance on algorithms.
Conclusion and Recommendations
More external validation is needed across diverse populations to meet fairness and generalizability. Ethical deployment of big data will require collaboration between clinicians, data scientists, and policymakers together with clear information about how data will be governed, and patients will be protected. Finally, AI tools are not to be used in lieu of clinical decision making; they are to be used in COPD and comorbid depression/Anxiety as the decision-support tools to achieve better personalized care.
REFERENCES
ADELOWO, O. F. (2023). LOGOTHERAPY AND COGNITIVE BEHAVIOURAL THERAPY IN THE TREATMENT OF DEPRESSIVE DISORDER AMONG STIGMATISED PEOPLE LIVING WITH HIV/AIDS IN OYO STATE, NIGERIA
Ayilya, B. L., & Nazeer, R. A. (2023). Epidemiological burden, risk factors, and recent therapeutic advances in chronic obstructive pulmonary disease. J Adv Biotechnol Exp Ther, 6, 109-122.
Feng, T., Li, P., Duan, R., & Jin, Z. (2025). Development and validation of a risk prediction model for depression in patients with chronic obstructive pulmonary disease. BMC Psychiatry, 25(1), 506. https://doi.org/10.1186/s12888-025-06913-1
Martínez-Gestoso, S., García-Sanz, M.-T., Carreira, J.-M., Salgado, F.-J., Calvo-Álvarez, U., Doval-Oubiña, L., Camba-Matos, S., Peleteiro-Pedraza, L., González-Pérez, M.-A., & Penela-Penela, P. (2022). Impact of anxiety and depression on the prognosis of copd exacerbations. BMC pulmonary medicine, 22(1), 169.
Moretta, P., Cavallo, N. D., Candia, C., Lanzillo, A., Marcuccio, G., Santangelo, G., Marcuccio, L., Ambrosino, P., & Maniscalco, M. (2024). Psychiatric Disorders in patients with chronic obstructive pulmonary disease: Clinical significance and treatment strategies. Journal of clinical medicine, 13(21), 6418.
Ostroff, J. L., & BCACP, B. (2021). Summarizing the 2021 updated GOLD guidelines for COPD. US Pharmacist, 46(7), 30-35.
Pozza, M., Navarin, N., Sakkalis, V., & Gabrielli, S. (2025). Artificial Intelligence Methods and Digital Intervention Strategies for Predicting and Managing Chronic Obstructive Pulmonary Disease Exacerbations: An Umbrella Review. Healthcare,
Siraj, R. A. (2025). COPD and comorbid mental health: addressing anxiety, and depression, and their clinical management. Medicina, 61(8), 1426.
Zhao, X., Wang, Y., Li, J., Liu, W., Yang, Y., Qiao, Y., Liao, J., Chen, M., Li, D., Wu, B., Huang, D., & Wu, D. (2025). A machine-learning-derived online prediction model for depression risk in COPD patients: A retrospective cohort study from CHARLS. Journal of Affective Disorders, 377, 284-293. https://doi.org/https://doi.org/10.1016/j.jad.2025.02.063