Background

Performance tests are important to characterize patient disabilities and functional changes. The Osteoarthritis Research Society International and others recommend the 30-second Chair Stand Test and Stair Climb Test, among others, as core tests that capture two distinct types of disability during activities of daily living. However, these two tests are limited by current protocols of testing in clinics. There is a need for an alternative that allows remote testing of functional capabilities during these tests in the osteoarthritis patient population.

Objective

Objectives are to (1) develop an app for testing the functionality of an iPhone's accelerometer and gravity sensor and (2) conduct a pilot study objectively evaluating the criterion validity and test-retest reliability of outcome variables obtained from these sensors during the 30-second Chair Stand Test and Stair Climb Test.

Methods

An iOS app was developed with data collection capabilities from the built-in iPhone accelerometer and gravity sensor tools and linked to Google Firebase. A total of 24 subjects performed the 30-second Chair Stand Test with an iPhone accelerometer collecting data and an external rater manually counting sit-to-stand repetitions. A total of 21 subjects performed the Stair Climb Test with an iPhone gravity sensor turned on and an external rater timing the duration of the test on a stopwatch. App data from Firebase were converted into graphical data and exported into MATLAB for data filtering. Multiple iterations of a data processing algorithm were used to increase robustness and accuracy. MATLAB-generated outcome variables were compared to the manually determined outcome variables of each test. Pearson's correlation coefficients (PCCs), Bland-Altman plots, intraclass correlation coefficients (ICCs), standard errors of measurement, and repeatability coefficients were generated to evaluate criterion validity, agreement, and test-retest reliability of iPhone sensor data against gold-standard manual measurements.

Results

App accelerometer data during the 30-second Chair Stand Test (PCC=.890) and gravity sensor data during the Stair Climb Test (PCC=.865) were highly correlated to gold-standard manual measurements. Greater than 95% of values on Bland-Altman plots comparing the manual data to the app data fell within the 95% limits of agreement. Strong intraclass correlation was found for trials of the 30-second Chair Stand Test (ICC=.968) and Stair Climb Test (ICC=.902). Standard errors of measurement for both tests were found to be within acceptable thresholds for MATLAB. Repeatability coefficients for the 30-second Chair Stand Test and Stair Climb Test were 0.629 and 1.20, respectively.

Conclusions

App-based performance testing of the 30-second Chair Stand Test and Stair Climb Test is valid and reliable, suggesting its applicability to future, larger-scale studies in the osteoarthritis patient population.

OBJECTIVE: Multiple sclerosis (MS) lesions are heterogeneous with regard to inflammation, demyelination, axonal injury, and neuronal loss. We previously developed a diffusion basis spectrum imaging (DBSI) technique to better address MS lesion heterogeneity. We hypothesized that the profiles of multiple DBSI metrics can identify lesion-defining patterns. Here we test this hypothesis by combining a deep learning algorithm using deep neural network (DNN) with DBSI and other imaging methods. METHODS: Thirty-eight MS patients were scanned with diffusion-weighted imaging, magnetization transfer imaging, and standard conventional MRI sequences (cMRI). A total of 499 regions of interest were identified on standard MRI and labeled as persistent black holes (PBH), persistent gray holes (PGH), acute black holes (ABH), acute gray holes (AGH), nonblack or gray holes (NBH), and normal appearing white matter (NAWM). DBSI, diffusion tensor imaging (DTI), and magnetization transfer ratio (MTR) were applied to the 43,261 imaging voxels extracted from these ROIs. The optimized DNN with 10 fully connected hidden layers was trained using the imaging metrics of the lesion subtypes and NAWM. RESULTS: Concordance, sensitivity, specificity, and accuracy were determined for the different imaging methods. DBSI-DNN derived lesion classification achieved 93.4% overall concordance with predetermined lesion types, compared with 80.2% for DTI-DNN model, 78.3% for MTR-DNN model, and 74.2% for cMRI-DNN model. DBSI-DNN also produced the highest specificity, sensitivity, and accuracy. CONCLUSIONS: DBSI-DNN improves the classification of different MS lesion subtypes, which could aid clinical decision making. The efficacy and efficiency of DBSI-DNN shows great promise for clinical applications in automatic MS lesion detection and classification.