This report summarizes the existing literature on the use of technology for the assessment of freezing of gait (FOG) as well as the use of technology to provide insights into the mechanisms of FOG in people with Parkinson's disease. Specifically, this work was carried out for the 3rd International Workshop on Freezing of Gait in Jerusalem in 2023. This review focuses on the most used technologies to quantitatively assess FOG in a laboratory environment and describes the technologies that hold promise for assessing FOG in daily life. Examples of implementation of machine learning algorithms are provided as well as algorithmic biases. Lastly, a standardized assessment using inertial measurement units during a clinical protocol is proposed and a 5-year outlook is discussed. We anticipate this review will help move the field forward in the coming years.

Freezing of gait (FOG) is a debilitating symptom of Parkinson's disease (PD), characterized by an absence or reduction in forward movement of the legs despite the intention to walk. Detecting FOG during free-living conditions presents significant challenges, particularly when using only inertial measurement unit (IMU) data, as it must be distinguished from voluntary stopping events that also feature reduced forward movement. Influences from stress and anxiety, measurable through galvanic skin response (GSR) and electrocardiogram (ECG), may assist in distinguishing FOG from normal gait and stopping. However, no study has investigated the fusion of IMU, GSR, and ECG for FOG detection. Therefore, this study introduced two methods: a two-step approach that first identified reduced forward movement segments using a Transformer-based model with IMU data, followed by an XGBoost model classifying these segments as FOG or stopping using IMU, GSR, and ECG features; and an end-to-end approach employing a multi-stage temporal convolutional network to directly classify FOG and stopping segments from IMU, GSR, and ECG data. Results showed that the two-step approach with all data modalities achieved an average F1 score of 0.728 and F1@50 of 0.725, while the end-to-end approach scored 0.771 and 0.759, respectively. However, no significant difference was found compared to using only IMU data in both approaches (p-values: 0.466 to 0.887). In conclusion, adding physiological data did not provide a statistically significant benefit in distinguishing between FOG and stopping. The limitations may be specific to GSR and ECG data, and may not generalize to other physiological modalities.

Purpose Physical activity through exercise helps to delay, prevent, or reverse functional decline in older adults (Bean et al., 2004). Recently, various technologies, including assistive robot coaches, have been developed to engage older adults in physical activities (Avioz-Sarig et al., 2021). Nevertheless, there is a lack of understanding regarding the specific needs, preferences, and potential barriers faced by older adults and their caregivers when interacting with this technology (Fasola and Mataric, 2012). This study is part of the interdisciplinary AI@WZC project, aimed at developing an AI-driven robotic exercise coach for older adults living in a residential care home. We aimed to acquire insights from residents and their caregivers to inform the design of this robotic coach. Method We organized two semi-structured focus groups with respectively five and four care home residents (age: 70–90 years old; gender: session 1 with three females and two males, session 2 with two females and two males; MMSE: 18-30) and two caregivers (one physical therapist, one nurse), adhering to the participatory design principles of the PERCEPT methodology (Bourazeri and Stumpf, 2018). During the first workshop, we gained insights into their technology use, physical challenges, and daily activities. This discussion aided the participants to co-create personas for a care home resident, a physical therapist, and a nurse. In the second workshop, these personas served as a basis to facilitate further discussions. Here, we asked participants to rank their preferred physical exercises related to training specific body parts (e.g., ankle, shoulder, neck), and inquired about their preferences related to functionalities the robot should have, with emphasis on the information that the robot should provide to them through its interface. Results and Discussion Participants have varying physical abilities and preferences regarding physical exercises they engage in. For example, care home residents use different mobility aids (e.g., walker, wheelchair), which impacts which exercises are suitable and safe. Therefore, it is necessary that caregivers can individually assess care home residents’ training progress with the robot, and make adjustments to ensure the exercises fit their individual needs. Flexibility is also important regarding the planning of the robotic coach’ visits. For example, the residents often engage in group or individual exercise sessions with a physical therapist. As such, participants voiced that training with the robot should not be scheduled too close to a session with a physical therapist. Furthermore, participants want to know in advance when an exercise session with the robot is planned, and want to have the freedom to cancel or reschedule such a session when it does not suit them, e.g., in case of unexpected family visitors, appointment with their doctor, or other activities. The design needs to take into account their limited experience with technology, while avoiding burdening caregivers with additional work load to manage the robot’s schedule. Regarding the interface of the robot, participants indicated that feedback primarily needs to be straightforward and functional, e.g., the robot needs to inform them about their progress during the training, letting them know how long it takes before they completed the exercise. Participants also expressed the wish not be interrupted during their training with the robot, meaning that non-functional distractions (e.g., visual embellishments, abundant feedback or encouragement) should be kept to a minimum. Summarizing these results, we identified two key user requirements based on the data collected during the two sessions. (1) The robotic coach needs to serve as an assessment tool, allowing caregivers to monitor functional decline and ensure the safety of exercise regimens. (2) It needs to enable personalization of exercise regimens according to individual preferences and physical limitations of care home residents, therefore allowing them to maintain agency regarding their exercise routines. At the same time, interaction between the robot and residents should be minimal, with a focus on functional feedback during exercises. In conclusion, this approach and the findings from this study can inspire and guide future researchers in developing assistive physical exercise technologies for use in residential care homes.

To gain insights into the impact of upper limb (UL) dysfunctions after breast cancer treatment, this study aimed to develop a temporal convolutional neural network (TCN) to detect functional daily UL use in breast cancer survivors using data from a wrist-worn accelerometer. A pre-existing dataset of 10 breast cancer survivors was used that contained raw 3-axis acceleration data and simultaneously recorded video data, captured during four daily life activities. The input of our TCN consists of a 3-axis acceleration sequence with a receptive field of 243 samples. The 4 ResNet TCN blocks perform dilated temporal convolutions with a kernel of size 3 and a dilation rate that increases by a factor of 3 after each iteration. Outcomes of interest were functional UL use (minutes) and percentage UL use. We found strong agreement between the video and predicted data for functional UL use (ICC = 0.975) and moderately strong agreement for %UL use (ICC = 0.794). The TCN model overestimated the functional UL use by 0.71 min and 3.06%. Model performance showed good accuracy, f1, and AUPRC scores (0.875, 0.909, 0.954, respectively). In conclusion, using wrist-worn accelerometer data, the TCN model effectively identified functional UL use in daily life among breast cancer survivors.

Background: Freezing of gait (FOG) is an episodic and highly disabling symptom of Parkinson’s Disease (PD). Traditionally, FOG assessment relies on time-consuming visual inspection of camera footage. Therefore, previous studies have proposed portable and automated solutions to annotate FOG. However, automated FOG assessment is challenging due to gait variability caused by medication effects and varying FOG-provoking tasks. Moreover, whether automated approaches can differentiate FOG from typical everyday movements, such as volitional stops, remains to be determined. To address these questions, we evaluated an automated FOG assessment model with deep learning (DL) based on inertial measurement units (IMUs). We assessed its performance trained on all standardized FOG-provoking tasks and medication states, as well as on specific tasks and medication states. Furthermore, we examined the effect of adding stopping periods on FOG detection performance. Methods: Twelve PD patients with self-reported FOG (mean age 69.33 ± 6.02 years) completed a FOG-provoking protocol, including timed-up-and-go and 360-degree turning-in-place tasks in On/Off dopaminergic medication states with/without volitional stopping. IMUs were attached to the pelvis and both sides of the tibia and talus. A temporal convolutional network (TCN) was used to detect FOG episodes. FOG severity was quantified by the percentage of time frozen (%TF) and the number of freezing episodes (#FOG). The agreement between the model-generated outcomes and the gold standard experts’ video annotation was assessed by the intra-class correlation coefficient (ICC). Results: For FOG assessment in trials without stopping, the agreement of our model was strong (ICC (%TF) = 0.92 [0.68, 0.98]; ICC(#FOG) = 0.95 [0.72, 0.99]). Models trained on a specific FOG-provoking task could not generalize to unseen tasks, while models trained on a specific medication state could generalize to unseen states. For assessment in trials with stopping, the agreement of our model was moderately strong (ICC (%TF) = 0.95 [0.73, 0.99]; ICC (#FOG) = 0.79 [0.46, 0.94]), but only when stopping was included in the training data. Conclusion: A TCN trained on IMU signals allows valid FOG assessment in trials with/without stops containing different medication states and FOG-provoking tasks. These results are encouraging and enable future work investigating automated FOG assessment during everyday life.

— Freezing of gait (FOG) is an episodic and highly disabling symptom of Parkinson’s disease (PD). Although described as a single phenomenon, FOG is heterogeneous and can express as different manifestations, such as trembling in place or complete akinesia. We aimed to analyze the efficacy of deep learning (DL) trained on inertial measurement unit data to classify FOG into both manifestations. We adapted and compared four state-of-the-art FOG detection algorithms for this task and investigated the advantages of incorporating a refinement model to address oversegmentation errors. We evaluated the model’s performance in distinguishing between trembling and akinesia, as well as other forms of movement cessation (e.g., stopping and sitting), against gold-standard video annotations. Experiments were conducted on a dataset of eighteen PD patients completing a FOG-provoking protocol in a gait laboratory. Results showed our model achieved an F1 score of 0.78 and segment F1@50 of 0.75 in detecting FOG manifestations. Assessment of FOG severity was strong for trembling (ICC=0.86, [0.66,0.95]) and moderately strong for akinesia (ICC=0.78, [0.51,0.91]). Importantly, our model successfully differentiated FOG from other forms of movement cessation during 360-degree turning-in-place tasks. In conclusion, our study demonstrates that DL can accurately assess different types of FOG manifestations, warranting further investigation in larger and more diverse verification cohorts.

The ability to identify and temporally segment fine-grained actions in motion capture sequences is crucial for applications in human movement analysis. Motion capture is typically performed with optical or inertial measurement systems, which encode human movement as a time series of human joint locations and orientations or their higher-order representations. State-of-the-art action segmentation approaches use multiple stages of temporal convolutions. The main idea is to generate an initial prediction with several layers of temporal convolutions and refine these predictions over multiple stages, also with temporal convolutions. Although these approaches capture long-term temporal patterns, the initial predictions do not adequately consider the spatial hierarchy among the human joints. To address this limitation, we recently introduced multi-stage spatial-temporal graph convolutional neural networks (MS-GCN). Our framework replaces the initial stage of temporal convolutions with spatial graph convolutions and dilated temporal convolutions, which better exploit the spatial configuration of the joints and their long-term temporal dynamics. Our framework was compared to four strong baselines on five tasks. Experimental results demonstrate that our framework is a strong baseline for skeleton-based action segmentation.

Continual Spatial-Temporal Graph Convolutional Network (CoST-GCN) is a continual inference optimization of ST-GCN, an established graph-based action classification method. It removes redundant computations in the ST-GCN classifier when applied as a sliding window over a continual stream of data for per-frame predictions. Despite the improvement of CoST-GCN we can only achieve a throughput of 5 fps on a representative edge platform (Raspberry Pi 4). We propose a hardware-driven optimization, termed RT-ST-GCN, which scales down the computational bottleneck of ST-GCN to achieve realtime predictions up to 50 fps. We study and compare the performance of our lightweight model against (Co)ST-GCN on the PKU-MMD continual action dataset. Despite an expected drop in framewise performance metrics, our model shows similar or better performance on key segmental metrics, a constant latency of 20 ms for any temporal kernel size and 3x decrease in memory usage.

Freezing of gait (FOG) is a common and severe symptom of Parkinson's disease (PD). Due to the complex underlying pathophysiology, FOG is difficult to assess, hampering further insight into this phenomenon. Inertial measurement units (IMUs) may enable FOG assessment during everyday life, but lack of standardization, e.g., the number and position of the IMUs, complicates an objective comparison of automatic FOG assessment algorithms. We propose a multi-stage temporal dilated convolutional model to automatically assess FOG based on IMU data. We collected simultaneous optical motion capture (MoCap) and IMU data of ten people with PD and FOG. We devised a simulation pipeline, i.e., generating IMU data from MoCap data, to objectively compare our approach to two state-of-The-Art FOG assessment models. The comparison was performed for five simulated IMU configurations, ranging from 1 to 7 IMUs. The results show that our approach outperforms the two state-of-The-Art methods on most of the simulated IMU configurations. The complete lower-body IMU setup of 7 IMUs (pelvis and both sides of the talus, tibia, and femur) enables the best FOG detection performance. Lastly, we show that our model trained by incorporating simulated IMU data enabled significantly improved FOG detection performance than our model trained only with real IMU data. In doing so, we demonstrate that retrospective MoCap datasets can be re-used to train expressive IMU-based FOG assessment models, reducing the required amount of dedicated and labor-intensive IMU data collection experiments.

Background: Freezing of gait (FOG) is a common and debilitating gait impairment in Parkinson’s disease. Further insight into this phenomenon is hampered by the difficulty to objectively assess FOG. To meet this clinical need, this paper proposes an automated motion-capture-based FOG assessment method driven by a novel deep neural network. Methods: Automated FOG assessment can be formulated as an action segmentation problem, where temporal models are tasked to recognize and temporally localize the FOG segments in untrimmed motion capture trials. This paper takes a closer look at the performance of state-of-the-art action segmentation models when tasked to automatically assess FOG. Furthermore, a novel deep neural network architecture is proposed that aims to better capture the spatial and temporal dependencies than the state-of-the-art baselines. The proposed network, termed multi-stage spatial-temporal graph convolutional network (MS-GCN), combines the spatial-temporal graph convolutional network (ST-GCN) and the multi-stage temporal convolutional network (MS-TCN). The ST-GCN captures the hierarchical spatial-temporal motion among the joints inherent to motion capture, while the multi-stage component reduces over-segmentation errors by refining the predictions over multiple stages. The proposed model was validated on a dataset of fourteen freezers, fourteen non-freezers, and fourteen healthy control subjects. Results: The experiments indicate that the proposed model outperforms four state-of-the-art baselines. Moreover, FOG outcomes derived from MS-GCN predictions had an excellent (r = 0.93 [0.87, 0.97]) and moderately strong (r = 0.75 [0.55, 0.87]) linear relationship with FOG outcomes derived from manual annotations. Conclusions: The proposed MS-GCN may provide an automated and objective alternative to labor-intensive clinician-based FOG assessment. Future work is now possible that aims to assess the generalization of MS-GCN to a larger and more varied verification cohort.

There are concerns about the stability of meropenem in plasma samples, even when frozen at −20 ^◦ C. Previous smaller studies suggested significant degradation of meropenem at −20 ^◦ C after 3–20 days. However, in several recent clinical studies, meropenem plasma samples were still stored at −20 ^◦ C, or the storage temperature and/or time were not mentioned in the paper. The aim of this study was to describe and model meropenem degradation in human plasma at −20 ^◦ C over 1 year. Stability of meropenem in human plasma at −20 ^◦ C was investigated at seven concentrations (0.44, 4.38, 17.5, 35.1, 52.6, 70.1, and 87.6 mg/L) representative for the range of relevant concentrations encountered in clinical practice. For each concentration, samples were stored for 0, 7, 14, 21, 28, 42, 56, 70, 84, 112, 140, 168, 196, 224, 252, 280, 308, 336, and 364 days at −20 ^◦ C before being transferred to −80 ^◦ C until analysis. Degradation was modeled using polynomial regression analysis and artificial neural network (ANN). Meropenem showed significant degradation over time in human plasma when stored at −20 ^◦ C. Degradation was present over the whole concentration range and increased with higher concentrations until a concentration of 35.1 mg/L. Both models showed accurate prediction of meropenem degradation. In conclusion, this study provides detailed insights into the concentration-dependent degradation of meropenem in human plasma stored at −20 ^◦ C over 1 year. Meropenem in human plasma is shown to be stable at least up to approximately 80 days when stored at −20 ^◦ C. The polynomial model allows calculating original meropenem concentrations in samples stored for a known period of time at −20 ^◦ C.