Machine Learning Models for Differentiating Causes of Dyspnea in Emergency Department Patients: A Retrospective Study
Justus Baarts, Jörg D. Leuppi, Maria Boesing, Giorgia Lüthi-Corridori · Journal of Multidisciplinary Healthcare · 2025
Background: Dyspnea is a frequent symptom in emergency departments (ED) with multifactorial causes, including cardiac and pulmonary conditions. Accurate and timely diagnosis is crucial to guide appropriate management and improve patient outcomes. Machine learning (ML) may aid this process. Methods: This retrospective study analyzed 787 adult patients presenting with dyspnea to the ED at Kantonsspital Baselland (Switzerland) in 2022. Clinical, laboratory, diagnostic data and final diagnoses were collected. ML models including decision trees, random forest and boosted decision trees were trained to classify dyspnea etiologies (cardiac vs respiratory) and predict final diagnoses. Performance metrics included accuracy, sensitivity, and specificity. Results: The most common diagnoses were decompensated heart failure (28.4%), pneumonia (26.4%), and COVID-19 (17%). Binary classification into cardiac vs respiratory causes achieved the highest performance (accuracy: 89.6% with boosted trees). Multiclass prediction of specific diagnoses yielded lower performance (accuracy: 36.8%). CRP, BNP, and cough emerged as key predictive features consistent with established clinical knowledge, supporting model interpretability. Comorbidities, though clinically relevant, showed limited predictive value. Conclusion: ML algorithms show promise in supporting triage by distinguishing broad etiological categories of dyspnea, such as respiratory versus cardiac origins. While the models demonstrate useful classification performance, their limited sensitivity for specific diagnoses underscores the need for larger, more diverse datasets and advanced modeling approaches. Plain Language Summary: People often come to the emergency department with shortness of breath, a symptom also called dyspnea. This can be caused by different health conditions, especially those affecting the heart or lungs. Quickly understanding the cause of dyspnea is important to ensure patients receive the right treatment. Our research team wanted to find out whether computer tools known as machine learning models could help doctors to better identify what is causing a person’s breathing problem. We analyze real-life data from 787 adults who came to the emergency department at Kantonsspital Baselland in Switzerland during 2022. For each person, we collected information like symptoms, lab results, and final diagnoses. We tested different types of machine learning models to classify whether the cause of a patient’s shortness of breath was related to the heart (for example, heart failure) or to the lungs (such as pneumonia or COVID-19). One model, called a boosted decision tree, was able to do this correctly almost 90% of the time. However, predicting the exact diagnosis was more challenging, with a lower accuracy. The most helpful indicators for the models were blood markers of inflammation, a heart-related hormone, and whether the patient had a cough. These results suggest that machine learning tools could support emergency teams by helping to rapidly distinguish between heart- and lung-related causes of shortness of breath. Further research using larger and more diverse patient groups will be needed to improve the accuracy and usefulness of these tools in clinical practice. Keywords: dyspnea, emergency department, ED, machine learning, ML, diagnosis prediction, boosted decision tree