A Novel Sensor-based Approach to Predict Adverse Outcomes in Chronic Obstructive Pulmonary Disease

Abstract

Chronic obstructive pulmonary disease (COPD) is a common progressive disease, which is the third leading cause of death among older adults in the United States. Over 80% of patients with COPD have at least one comorbid chronic condition, and 25% are frail. Adverse outcome prediction using function measures reflective of total burden of COPD patients is important because it identifies vulnerable patients and helps to enhance bronchodilator therapies and specialized COPD treatment strategies including pharmacotherapy, rehabilitation, exercise programs, and measures to consolidate care in this complex population. COPD diagnosis, prognosis, and pharmacologic and non-pharmacologic treatment are assessed using spirometry, with low forced expiratory volume in one second (FEV1) predicting worse health outcomes. While FEV1 is related to mortality, it is not sensitive in “adverse outcome prediction” to identify vulnerable patients.Assessment of functional capacity (reflective of the geriatric syndromes of dynapenia and clinical frailty), in combination with pulmonary function testing (PFT) can help to predict COPD health outcomes. The 6-minute walk distance (6MWD) test is commonly used to assess functional capacity in COPD patients and has been shown to predict mortality better than PFTs. While the 6MWD has good reliability and validity, it is time-consuming, burdensome in clinical settings, or simply not feasible for some patients, especially elders with mobility impairments. For these reasons, an alternative objective, quick, and simple approach for assessing functional capacity in COPD would be beneficial to risk stratify, predict outcomes, and direct appropriate care. Previously, we developed an upper extremity function (UEF) test for frailty assessment, and we developed a frailty index based on the kinematic information that we obtain from the arm motion. Our previous investigations showed a correlation between the UEF frailty index and COPD 12 adverse outcomes. However, the frailty score lacks direct assessment of muscle dysfunction. Studies showed that dynapenia (age-related loss of muscle performance) is related to limb muscle dysfunction, and is associated with reduced exercise tolerance, quality of life, and mortality in COPD. We extended the UEF approach by developing a two degree-of-freedom subject-specific computational muscle model to simulate elbow motions and predict muscle forces during the UEF test. We used optimization approach and recruited entropy assisted cost function to predict muscle forces that simulate the patients’ elbow motions. We collected EMG signals of biceps and triceps to verify the model. Muscle parameters were extracted from muscle forces to study muscle dysfunction in COPD patients. Our model was able to predict muscle co-contraction and muscle force ratio during arm flexion extension motions without using EMG data. Along with kinematic parameters, we extracted muscle model parameters to more accurately study muscle dysfunction in COPD patients. In Order to train and test the UEF model for adverse outcome prediction, we recruited 192 admitted COPD patients and asked them to perform the UEF test while they were in the hospital. In-hospital and longitudinal outcomes were recorded by following up the patients for 90 days. We used elastic- net regularization for selecting optimum feature sets (kinematics and muscle model parameters) in combination with machine learning (i.e., support vector machine (SVM), K-Nearest Neighbors (KNN), and Logistic regression) to predict adverse outcomes. Current findings suggested that using UEF approach and studying kinematic and muscle dysfunction based on the muscle model may provide an efficient method for risk stratifying older adults with COPD with accuracy higher than available tools (e.g., clinical frailty score and COPD assessment). This approach is quick, 13 objective, and more feasible compared to 6MWD test to predict adverse outcomes in patients with COPD.