Machine learning-based detection of acute psychosocial stress from body posture and movements.

Investigating acute stress responses is crucial to understanding the underlying mechanisms of stress. Current stress assessment methods include self-reports that can be biased and biomarkers that are often based on complex laboratory procedures. A promising additional modality for stress assessment might be the observation of body movements, which are affected by negative emotions and threatening situations. In this paper, we investigated the relationship between acute psychosocial stress induction and body posture and movements. We collected motion data from N = 59 individuals over two studies (Pilot Study: N = 20, Main Study: N = 39) using inertial measurement unit (IMU)-based motion capture suits. In both studies, individuals underwent the Trier Social Stress Test (TSST) and a stress-free control condition (friendly-TSST; f-TSST) in randomized order. Our results show that acute stress induction leads to a reproducible freezing behavior, characterized by less overall motion as well as more and longer periods of no movement. Based on these data, we trained machine learning pipelines to detect acute stress solely from movement information, achieving an accuracy of 75.0 ± 17.7 % (Pilot Study) and 73.4 ± 7.7 % (Main Study). This, for the first time, suggests that body posture and movements can be used to detect whether individuals are exposed to acute psychosocial stress. While more studies are needed to further validate our approach, we are convinced that motion information can be a valuable extension to the existing biomarkers and can help to obtain a more holistic picture of the human stress response. Our work is the first to systematically explore the use of full-body body posture and movement to gain novel insights into the human stress response and its effects on the body and mind.

The effects of an individualized smartphone-based exercise program on self-defined motor tasks in Parkinson's disease: a long-term feasibility study.

BACKGROUND: Exercise therapy is considered effective for the treatment of motor impairment in patients with Parkinson's disease (PD). During the COVID-19 pandemic, training sessions were cancelled and the implementation of telerehabilitation concepts became a promising solution. The aim of this controlled interventional feasibility study was to evaluate the long-term acceptance and to explore initial effectiveness of a digital, home-based, high-frequency exercise program for PD patients. Training effects were assessed using patient-reported outcome measures combined with sensor-based and clinical scores. METHODS: 16 PD patients (smartphone group, SG) completed a home-based, individualized training program over 6-8 months using a smartphone app, remotely supervised by a therapist, and tailored to the patient's motor impairments and capacity. A control group (CG, n = 16) received medical treatment without participating in digital exercise training. The usability of the app was validated using System Usability Scale (SUS) and User Version of the Mobile Application Rating Scale (uMARS). Outcome measures included among others Unified Parkinson Disease Rating Scale, part III (UPDRS-III), sensor-based gait parameters derived from standardized gait tests, Parkinson's Disease Questionnaire (PDQ-39), and patient-defined motor activities of daily life (M-ADL). RESULTS: Exercise frequency of 74.5% demonstrated high adherence in this cohort. The application obtained 84% in SUS and more than 3.5/5 points in each subcategory of uMARS, indicating excellent usability. The individually assessed additional benefit showed at least 6 out of 10 points (Mean = 8.2 ± 1.3). From a clinical perspective, patient-defined M-ADL improved for 10 out of 16 patients by 15.5% after the training period. The results of the UPDRS-III remained stable in the SG while worsening in the CG by 3.1 points (24%). The PDQ-39 score worsened over 6-8 months by 83% (SG) and 59% (CG) but the subsection mobility showed a smaller decline in the SG (3%) compared to the CG (77%) without reaching significance level for all outcomes. Sensor-based gait parameters remained constant in both groups. CONCLUSIONS: Long-term training over 6-8 months with the app is considered feasible and acceptable, representing a cost-effective, individualized approach to complement dopaminergic treatment. This study indicates that personalized, digital, high-frequency training leads to benefits in motor sections of ADL and Quality of Life.

Validity of an inertial sensor-based system for the assessment of spatio-temporal parameters in people with multiple sclerosis.

Background: Gait variability in people with multiple sclerosis (PwMS) reflects disease progression or may be used to evaluate treatment response. To date, marker-based camera systems are considered as gold standard to analyze gait impairment in PwMS. These systems might provide reliable data but are limited to a restricted laboratory setting and require knowledge, time, and cost to correctly interpret gait parameters. Inertial mobile sensors might be a user-friendly, environment- and examiner-independent alternative. The purpose of this study was to evaluate the validity of an inertial sensor-based gait analysis system in PwMS compared to a marker-based camera system. Methods: A sample N = 39 PwMS and N = 19 healthy participants were requested to repeatedly walk a defined distance at three different self-selected walking speeds (normal, fast, slow). To measure spatio-temporal gait parameters (i.e., walking speed, stride time, stride length, the duration of the stance and swing phase as well as max toe clearance), an inertial sensor system as well as a marker-based camera system were used simultaneously. Results: All gait parameters highly correlated between both systems (r > 0.84) with low errors. No bias was detected for stride time. Stance time was marginally overestimated (bias = -0.02 ± 0.03 s) and gait speed (bias = 0.03 ± 0.05 m/s), swing time (bias = 0.02 ± 0.02 s), stride length (0.04 ± 0.06 m), and max toe clearance (bias = 1.88 ± 2.35 cm) were slightly underestimated by the inertial sensors. Discussion: The inertial sensor-based system captured appropriately all examined gait parameters in comparison to a gold standard marker-based camera system. Stride time presented an excellent agreement. Furthermore, stride length and velocity presented also low errors. Whereas for stance and swing time, marginally worse results were observed.