User login
DIFFUSE LUNG DISEASE AND LUNG TRANSPLANT NETWORK
Pulmonary Physiology and Rehabilitation Section
Several studies have explored automation of CPET interpretation, the most notable of which utilized machine learning.1
Recently, Schwendinger et al. investigated the ability of machine learning algorithms to not only categorize (pulmonary-vascular, mechanical-ventilatory, cardiocirculatory, and muscular), but also assign severity scores (0-6) to exercise limitations found in a group of 200 CPETs performed on adult patients referred to a lung clinic in Germany.2 Decision trees were constructed for each of the limitation categories by identifying variables with the lowest Root Mean Square Error (RMSE), which were comparable to agreement within expert interpretations. Combining decision trees allowed for a more comprehensive analysis with identification of multiple abnormalities in the same test.
A major limitation to the study is limited applicability to general patient populations without suspected lung disease. This bias is reflected in the decision tree for cardiovascular limitation that relied on VO2 peak and FEV1 alone. The authors were unable to construct a decision tree for muscular limitations due to a lack of identified cases.
Overall, these results suggest that refinement of machine learning algorithms built with larger heterogeneous data sets and expert interpretation can make CPETs accessible to the nonexpert clinician as long as test quality can be replicated across centers.
–Joseph Russo, MD
Fellow-in-Training
– Fatima Zeba, MD
Member-at-Large
References
1. Portella JJ, Andonian BJ, Brown DE, et al. Using machine learning to identify organ system specific limitations to exercise via cardiopulmonary exercise testing. IEEE J Biomed Health Inform. 2022;26(8):4228-4237.
2. Schwendinger F, Biehler AK, Nagy-Huber M, et al. Using machine learning-based algorithms to identify and quantify exercise limitations in clinical practice: are we there yet? Med Sci Sports Exerc. 2024;56(2):159-169.
DIFFUSE LUNG DISEASE AND LUNG TRANSPLANT NETWORK
Pulmonary Physiology and Rehabilitation Section
Several studies have explored automation of CPET interpretation, the most notable of which utilized machine learning.1
Recently, Schwendinger et al. investigated the ability of machine learning algorithms to not only categorize (pulmonary-vascular, mechanical-ventilatory, cardiocirculatory, and muscular), but also assign severity scores (0-6) to exercise limitations found in a group of 200 CPETs performed on adult patients referred to a lung clinic in Germany.2 Decision trees were constructed for each of the limitation categories by identifying variables with the lowest Root Mean Square Error (RMSE), which were comparable to agreement within expert interpretations. Combining decision trees allowed for a more comprehensive analysis with identification of multiple abnormalities in the same test.
A major limitation to the study is limited applicability to general patient populations without suspected lung disease. This bias is reflected in the decision tree for cardiovascular limitation that relied on VO2 peak and FEV1 alone. The authors were unable to construct a decision tree for muscular limitations due to a lack of identified cases.
Overall, these results suggest that refinement of machine learning algorithms built with larger heterogeneous data sets and expert interpretation can make CPETs accessible to the nonexpert clinician as long as test quality can be replicated across centers.
–Joseph Russo, MD
Fellow-in-Training
– Fatima Zeba, MD
Member-at-Large
References
1. Portella JJ, Andonian BJ, Brown DE, et al. Using machine learning to identify organ system specific limitations to exercise via cardiopulmonary exercise testing. IEEE J Biomed Health Inform. 2022;26(8):4228-4237.
2. Schwendinger F, Biehler AK, Nagy-Huber M, et al. Using machine learning-based algorithms to identify and quantify exercise limitations in clinical practice: are we there yet? Med Sci Sports Exerc. 2024;56(2):159-169.
DIFFUSE LUNG DISEASE AND LUNG TRANSPLANT NETWORK
Pulmonary Physiology and Rehabilitation Section
Several studies have explored automation of CPET interpretation, the most notable of which utilized machine learning.1
Recently, Schwendinger et al. investigated the ability of machine learning algorithms to not only categorize (pulmonary-vascular, mechanical-ventilatory, cardiocirculatory, and muscular), but also assign severity scores (0-6) to exercise limitations found in a group of 200 CPETs performed on adult patients referred to a lung clinic in Germany.2 Decision trees were constructed for each of the limitation categories by identifying variables with the lowest Root Mean Square Error (RMSE), which were comparable to agreement within expert interpretations. Combining decision trees allowed for a more comprehensive analysis with identification of multiple abnormalities in the same test.
A major limitation to the study is limited applicability to general patient populations without suspected lung disease. This bias is reflected in the decision tree for cardiovascular limitation that relied on VO2 peak and FEV1 alone. The authors were unable to construct a decision tree for muscular limitations due to a lack of identified cases.
Overall, these results suggest that refinement of machine learning algorithms built with larger heterogeneous data sets and expert interpretation can make CPETs accessible to the nonexpert clinician as long as test quality can be replicated across centers.
–Joseph Russo, MD
Fellow-in-Training
– Fatima Zeba, MD
Member-at-Large
References
1. Portella JJ, Andonian BJ, Brown DE, et al. Using machine learning to identify organ system specific limitations to exercise via cardiopulmonary exercise testing. IEEE J Biomed Health Inform. 2022;26(8):4228-4237.
2. Schwendinger F, Biehler AK, Nagy-Huber M, et al. Using machine learning-based algorithms to identify and quantify exercise limitations in clinical practice: are we there yet? Med Sci Sports Exerc. 2024;56(2):159-169.