A Deep Learning Approach for Automatic and Objective Grading of the Motor Impairment Severity in Parkinson's Disease for Use in Tele-Assessments.

Wearable sensors provide a tool for at-home monitoring of motor impairment progression in neurological conditions such as Parkinson's disease (PD). This study examined the ability of deep learning approaches to grade the motor impairment severity in a modified version of the Movement Disorders Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) using low-cost wearable sensors. We hypothesized that expanding training datasets with motion data from healthy older adults (HOAs) and initializing classifiers with weights learned from unsupervised pre-training would lead to an improvement in performance when classifying lower vs. higher motor impairment relative to a baseline deep learning model (XceptionTime). This study evaluated the change in classification performance after using expanded training datasets with HOAs and transferring weights from unsupervised pre-training compared to a baseline deep learning model (XceptionTime) using both upper extremity (finger tapping, hand movements, and pronation-supination movements of the hands) and lower extremity (toe tapping and leg agility) tasks consistent with the MDS-UPDRS. Overall, we found a 12.2% improvement in accuracy after expanding the training dataset and pre-training using max-vote inference on hand movement tasks. Moreover, we found that the classification performance improves for every task except toe tapping after the addition of HOA training data. These findings suggest that learning from HOA motion data can implicitly improve the representations of PD motion data for the purposes of motor impairment classification. Further, our results suggest that unsupervised pre-training can improve the performance of motor impairment classifiers without any additional annotated PD data, which may provide a viable solution for a widely deployable telemedicine solution.

Corrigendum to "Feasibility of virtual low-cost quantitative continuous measurement of movements in the extremities of people with Parkinson's disease" [MethodsX 11 (2023) 102230].

[This corrects the article DOI: 10.1016/j.mex.2023.102230.].

Dataset of quality assurance measurements of rhythmic movements.

A low-cost quantitative structured office measurement of movements in the extremities of people with Parkinson's disease [1,2] was performed on participants with Parkinson's disease and multiple system atrophy as well as age- and sex-matched healthy participants with typical development. Participants underwent twelve videotaped procedures rated by a trained examiner while connected to four accelerometers [1,2] generating a trace of the three location dimensions expressed as spreadsheets [3,4]. The signals of the five repetitive motion items (3.4 Finger tapping, 3.5 Hand movements, 3.6 Pronation-supination movements of hands, 3.7 Toe tapping, and 3.8 Leg agility) [1] underwent processing to fast Fourier [5] and amor and bump continuous wavelet transforms [6], [7], [8], [9], [10], [11], [12], [13]. Images of the signals and their transforms [4], [5], [6] of the five repetitive tasks of each participant were randomly expressed as panels on an electronic framework for rating by 35 trained examiners who did not know the source of the original output [14]. The team of international raters completed ratings of the signals and their transforms independently using criteria like the scoring systems for live assessments of movements in human participants [1,2]. The raters scored signals and transforms for deficits in the sustained performance of rhythmic movements (interruptions, slowing, and amplitude decrements) often observed in people with Parkinson's disease [15], [16], [17], [18], [19], [20]. Raters were first presented the images of the signals and transforms of a man with multiple system atrophy as a test and a retest in a different random order. After the raters completed the assessments of the man with multiple system atrophy, they were presented random test and retest panels of the images of signals and transforms of ten participants with Parkinson's disease who completed a single rating session. After the raters completed the assessments of the participants with Parkinson's disease who completed one set of ratings, they were presented random test and retest panels of the images of signals and transforms of (A) ten participants with Parkinson's disease and (B) eight age- and sex-match healthy participants with typical development who completed two rating session separated by a month or more [15], [16], [17], [18], [19], [20]. The data provide a framework for further analysis of the acquired information. Additionally, the data provide a template for the construction of electronic frameworks for the remote analysis by trained raters of signals and transforms of rhythmic processes to verify that the systems are operating smoothly without interruptions or changes in frequency and amplitude. Thus, the data provide the foundations to construct electronic frameworks for the virtual quality assurance of a vast spectrum of rhythmic processes. The dataset is a suitable template for solving unsupervised and supervised machine learning algorithms. Readers may utilize this procedure to assure the quality of rhythmic processes by confirming the absence of deviations in rate and rhythm. Thus, this procedure provides the means to confirm the quality of the vast spectrum of rhythmic processes.

Feasibility of virtual low-cost quantitative continuous measurement of movements in the extremities of people with Parkinson's disease.

A low-cost quantitative continuous measurement of movements in the extremities of people with Parkinson's disease, a structured motor assessment administered by a trained examiner to a patient physically present in the same room, utilizes sensors to generate output to facilitate the evaluation of the patient. However, motor assessments with the patient and the examiner in the same room may not be feasible due to distances between the patient and the examiner and the risk of transmission of infections between the patient and the examiner. Therefore, we propose a protocol for the remote assessment by examiners in different locations of both (A) videos of patients recorded during in-person motor assessments and (B) live virtual assessments of patients in different locations from examiners. The proposed procedure provides a framework for providers, investigators, and patients in vastly diverse locations to conduct optimal motor assessments required to develop treatment plans utilizing precision medicine tailored to the specific needs of each individual patient. The proposed protocol generates the foundation for providers to remotely perform structured motor assessments necessary for optimal diagnosis and treatment of people with Parkinson's disease and related conditions.

Classification of extremity movements by visual observation of signals and their transforms.

A low-cost quantitative continuous measurement of movements utilizes accelerometers to generate signal outputs to precisely record the positions of extremities during the performance of movements. This procedure can readily be accomplished with inexpensive materials constructed indivisuals throughout the world. The proposed protocol provides the framework for trained raters to assess the signal outputs by visual observation to generate objective measurements like the measurements of the actual movements. Expert raters can then remotely give quantitative suggestions for providers in underserved regions to utilize precision medicine to develop optimal treatment plans tailored to the specific needs of each individual. The proposed protocol lays the foundations for experts located in tertiary centers to provide optimal assessments of signal outputs generated remotely in underserved regions. This protocol provides the means to address gaps in current research including the dearth of objective measurements of movements utilizing automatic intelligence and machine learning to accurately and precisely analyze movement assessments. Future research will include the development of robotic tools to perform assessments and analyses of the movements of human beings to enhance the conduct of movement evaluations of people with Parkinson's disease and related conditions to apply precision medicine for optimal diagnostic and therapeutic interventions.

Trunk repositioning errors are increased in balance-impaired older adults.

BACKGROUND: Controlling the flexing trunk is critical in recovering from a loss of balance and avoiding a fall. To investigate the relationship between trunk control and balance in older adults, we measured trunk repositioning accuracy in young and balance-impaired and unimpaired older adults. METHODS: Young adults (N = 8, mean age 24.3 years) and two groups of community-dwelling older adults defined by unipedal stance time (UST)-a balance-unimpaired group (UST > 30 seconds, N = 7, mean age 73.9 years) and a balance-impaired group (UST < 5 seconds, N = 8, mean age 79.6 years)-were tested in standing trunk control ability by reproducing a approximately 30 degrees trunk flexion angle under three visual-surface conditions: eyes opened and closed on the floor, and eyes opened on foam. Errors in reproducing the angle were defined as trunk repositioning errors (TREs). Clinical measures related to balance, trunk extensor strength, and self-reported disability were obtained. RESULTS: TREs were significantly greater in the balance-impaired group than in the other groups, even when controlling for trunk extensor strength and body mass. In older adults, there were significant correlations between TREs and three clinical measures of balance and fall risk, UST and maximum step length (-0.65 to -0.75), and Timed Up & Go score (0.55), and between TREs and age (0.63-0.76). In each group TREs were similar under the three visual-surface conditions. Test-retest reliability for TREs was good to excellent (intraclass correlation coefficients > or =0.74). CONCLUSIONS: Older balance-impaired adults have larger TREs, and thus poorer trunk control, than do balance-unimpaired older individuals. TREs are reliable and valid measures of underlying balance impairment in older adults, and may eventually prove to be useful in predicting the ability to recover from losses of balance and to avoid falls.