Developing and validating an accelerometer-based algorithm with machine learning to classify physical activity after acquired brain injury.

Purpose: To develop and validate an accelerometer-based algorithm classifying physical activity in people with acquired brain injury (ABI) in a laboratory setting resembling a real home environment.Materials and methods: A development and validation study was performed. Eleven healthy participants and 25 patients with ABI performed a protocol of transfers and ambulating activities. Activity measurements were performed with accelerometers and with thermal video camera as gold standard reference. A machine learning-based algorithm classifying specific physical activities from the accelerometer data was developed and cross-validated in a training sample of 11 healthy participants. Criterion validity of the algorithm was established in 3 models classifying the same protocol of activities in people with ABI.Results: Modeled on data from 11 healthy and 15 participants with ABI, the algorithm had a good precision for classifying transfers and ambulating activities in data from 10 participants with ABI. The weighted sensitivity for all activities was 89.3% (88.3-90.4%) and the weighted positive predictive value was 89.7% (88.7-90.7%). The algorithm differentiated between lying and sitting activities.Conclusion: An algorithm to classify physical activities in populations with ABI was developed and its criterion validity established. Further testing of precision in home settings with continuous activity monitoring is warranted.

Exploring Physical Activity During the Discharge Transition Phase in People With Acquired Brain Injury-An Observational Study.

Objective: To explore physical activity trajectories during the discharge transition phase after in-hospital rehabilitation after acquired brain injury (ABI). Design: A cross-sectional observational study. Setting: Transition from an in-hospital rehabilitation center to community-based living. Participants: Independently walking patients with ABI (n=10) who were ready for discharge. Interventions: Not applicable. Main Outcome Measures: Two weeks of physically active time continuously monitored with an accelerometer and classified by a machine learning algorithm summed as daily average and total active time for each participant and classified into standing, walking, running, bike riding, stair climbing, ambulation, and sedentary time. Physical activity trajectories showing the total daily active time for all participants were inspected before and after discharge, and the average active time per participant was plotted against self-reported scores of potentially explanatory factors. Results: Average total physically active time was 5:49 hours (range 4:26-7:13 hours). Average daily physically active time for participants appeared to be related to functional independence measure sub scores, fatigue, and pre-morbid physical activity level. Individual physical activity trajectories showed a decreased walking activity after discharge, which increased again after 1-2 days. Conclusions: Daily total physically active time among participants was higher than expected. Factors expectedly related to physical activity trajectories in the discharge transition phase were explored and showed some relation to functional scores.

Improved active participation among patients with severe acquired brain injury in early rehabilitation: Implementation of a professional competence development intervention.

BACKGROUND: In 2019, an educational programme was implemented in a sub-acute in-hospital neurorehabilitation clinic for patients with severe acquired brain injury (sABI). The programme was initiated to enhance staff competencies related to identifying and improving active participation among patients with sABI. OBJECTIVE: The purpose was to evaluate the implementation effectiveness of the educational programme. METHODS: Mixed methods were chosen to assess implementation effectiveness as perceived by staff and patients. RESULTS: A survey of the professional's experience showed an increase in perceived competence after each completed seminar and from before the first seminar to after the last completed seminar. These results were confirmed and elaborated through staff focus group interviews. The proportion of patients achieving active participation increased from 45% before to 75% after implementation (six of eight patients). CONCLUSION: Exploring the implementation effectiveness of the educational programme seemed clinically valuable and showed a promising and probable effect of an implementation process.

Monitoring Physical Behavior in Rehabilitation Using a Machine Learning-Based Algorithm for Thigh-Mounted Accelerometers: Development and Validation Study.

BACKGROUND: Physical activity is emerging as an outcome measure. Accelerometers have become an important tool in monitoring physical behavior, and newer analytical approaches of recognition methods increase the degree of details. Many studies have achieved high performance in the classification of physical behaviors through the use of multiple wearable sensors; however, multiple wearables can be impractical and lower compliance. OBJECTIVE: The aim of this study was to develop and validate an algorithm for classifying several daily physical behaviors using a single thigh-mounted accelerometer and a supervised machine-learning scheme. METHODS: We collected training data by adding the behavior classes-running, cycling, stair climbing, wheelchair ambulation, and vehicle driving-to an existing algorithm with the classes of sitting, lying, standing, walking, and transitioning. After combining the training data, we used a random forest learning scheme for model development. We validated the algorithm through a simulated free-living procedure using chest-mounted cameras for establishing the ground truth. Furthermore, we adjusted our algorithm and compared the performance with an existing algorithm based on vector thresholds. RESULTS: We developed an algorithm to classify 11 physical behaviors relevant for rehabilitation. In the simulated free-living validation, the performance of the algorithm decreased to 57% as an average for the 11 classes (F-measure). After merging classes into sedentary behavior, standing, walking, running, and cycling, the result revealed high performance in comparison to both the ground truth and the existing algorithm. CONCLUSIONS: Using a single thigh-mounted accelerometer, we obtained high classification levels within specific behaviors. The behaviors classified with high levels of performance mostly occur in populations with higher levels of functioning. Further development should aim at describing behaviors within populations with lower levels of functioning.

Mind the gap. How job task distributions of health professional developers constitute evidence-based practice.

BACKGROUND: Development of clinical practice at a Danish neurorehabilitation centre was delegated to a group of health professional developers. Their job function lacked conceptual foundation, and it was unclear how their working tasks contributed to evidence-based practice. OBJECTIVE: Conceptual clarification of the job function and pattern analysis of activity distributions for health professional developers. METHODS: Health professional developers kept continuous time geographical diaries for two weeks. Meaningful categories were subtracted through content analysis. Patterns were analysed within activity distributions with regards to evidence-based practice. RESULTS: A total of 213 diaries were collected from 21 health professional developers of three professions (physiotherapists, occupational therapists and nurses). Each participant reported 6-13 workdays (median 10 days). Eleven main categories of work tasks emerged with 42 subcategories. Overall, 7% of total time reported was spent on external knowledge, with minimal variation between professions and contractual time allocation. CONCLUSION: Conceptual clarification of work tasks was established for health professional developers. Their work activity distributions contributed mainly to maintenance of existing level of professional knowledge rather than to implementation of new knowledge, which did not fulfil the intended responsibility for development of evidence-based practice. Educational competence boost and data-driven change of organisation structure was recommended.

Using Routinely Gathered Clinical Data to Develop a Prognostic Online Tool for Decannulation in Subjects With Acquired Brain Injury.

BACKGROUND: Clinicians are often required to provide a qualified guess on the probability of decannulation in estimating patients' rehabilitation potential and relaying information about prognosis to patients and next of kin. The objective of this study was to use routinely gathered clinical data to develop a prognostic model of time to decannulation in subjects with acquired brain injury, for direct implementation in clinical practice. METHODS: Data from a large cohort including 574 tracheostomized subjects admitted for neurorehabilitation were analyzed using discrete time-to-event analysis with logit-link. Within this model, a reference hazard function was modeled using restricted cubic splines, and estimates were presented using odds ratios (95% CIs). RESULTS: A total of 411 subjects (72%) were decannulated within a median of 27 d (interquartile range 16-49) at the rehabilitation hospital. The prognostic model for decannulation included age, diagnosis, days from injury until admission for rehabilitation, swallowing, and overall functional level measured with the Early Functional Abilities score. Among these, the strongest predictors for decannulation were age and a combination of overall functional abilities combined with swallowing ability. CONCLUSIONS: A prognostic model for decannulation was developed using routinely gathered clinical data. Based on the model, an online graphical user interface was applied, in which the probability of decannulation within x days is calculated along with the statistical uncertainty of the probability. Furthermore, a layman's interpretation is provided. The online tool was directly implemented in clinical practice at the rehabilitation hospital, and is available through this link: (http://www.hospitalsenhedmidt.dk/regionshospitalet-hammel/research-unit/Prognosissoftware/).