1
|
Klaver EC, Heijink IB, Silvestri G, van Vugt JPP, Janssen S, Nonnekes J, van Wezel RJA, Tjepkema-Cloostermans MC. Comparison of state-of-the-art deep learning architectures for detection of freezing of gait in Parkinson's disease. Front Neurol 2023; 14:1306129. [PMID: 38178885 PMCID: PMC10764416 DOI: 10.3389/fneur.2023.1306129] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/03/2023] [Accepted: 11/21/2023] [Indexed: 01/06/2024] Open
Abstract
Introduction Freezing of gait (FOG) is one of the most debilitating motor symptoms experienced by patients with Parkinson's disease (PD). FOG detection is possible using acceleration data from wearable sensors, and a convolutional neural network (CNN) is often used to determine the presence of FOG epochs. We compared the performance of a standard CNN for the detection of FOG with two more complex networks, which are well suited for time series data, the MiniRocket and the InceptionTime. Methods We combined acceleration data of people with PD across four studies. The final data set was split into a training (80%) and hold-out test (20%) set. A fifth study was included as an unseen test set. The data were windowed (2 s) and five-fold cross-validation was applied. The CNN, MiniRocket, and InceptionTime models were evaluated using a receiver operating characteristic (ROC) curve and its area under the curve (AUC). Multiple sensor configurations were evaluated for the best model. The geometric mean was subsequently calculated to select the optimal threshold. The selected model and threshold were evaluated on the hold-out and unseen test set. Results A total of 70 participants (23.7 h, 9% FOG) were included in this study for training and testing, and in addition, 10 participants provided an unseen test set (2.4 h, 11% FOG). The CNN performed best (AUC = 0.86) in comparison to the InceptionTime (AUC = 0.82) and MiniRocket (AUC = 0.76) models. For the CNN, we found a similar performance for a seven-sensor configuration (lumbar, upper and lower legs and feet; AUC = 0.86), six-sensor configuration (upper and lower legs and feet; AUC = 0.87), and two-sensor configuration (lower legs; AUC = 0.86). The optimal threshold of 0.45 resulted in a sensitivity of 77% and a specificity of 58% for the hold-out set (AUC = 0.72), and a sensitivity of 85% and a specificity of 68% for the unseen test set (AUC = 0.90). Conclusion We confirmed that deep learning can be used to detect FOG in a large, heterogeneous dataset. The CNN model outperformed more complex networks. This model could be employed in future personalized interventions, with the ultimate goal of using automated FOG detection to trigger real-time cues to alleviate FOG in daily life.
Collapse
Affiliation(s)
- Emilie Charlotte Klaver
- Department of Neurology and Clinical Neurophysiology, Medical Spectrum Twente, Enschede, Netherlands
- Department of Neurobiology, Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, Netherlands
| | - Irene B. Heijink
- Department of Neurology and Clinical Neurophysiology, Medical Spectrum Twente, Enschede, Netherlands
| | - Gianluigi Silvestri
- Department of Neurobiology, Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, Netherlands
- OnePlanet Research Center imec-the Netherlands, Wageningen, Netherlands
| | - Jeroen P. P. van Vugt
- Department of Neurology and Clinical Neurophysiology, Medical Spectrum Twente, Enschede, Netherlands
| | - Sabine Janssen
- Department of Rehabilitation, Centre of Expertise for Parkinson and Movement Disorders, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Centre, Nijmegen, Netherlands
- Department of Biomedical Signals and Systems, MedTech Centre, University of Twente, Enschede, Netherlands
- Department of Neurology, Anna Hospital, Geldrop, Netherlands
| | - Jorik Nonnekes
- Department of Rehabilitation, Centre of Expertise for Parkinson and Movement Disorders, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Centre, Nijmegen, Netherlands
- Department of Rehabilitation, Sint Maartenskliniek, Nijmegen, Netherlands
| | - Richard J. A. van Wezel
- Department of Neurobiology, Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, Netherlands
- Department of Biomedical Signals and Systems, MedTech Centre, University of Twente, Enschede, Netherlands
| | - Marleen C. Tjepkema-Cloostermans
- Department of Neurology and Clinical Neurophysiology, Medical Spectrum Twente, Enschede, Netherlands
- Department of Clinical Neurophysiology, MedTech Centre, University of Twente, Enschede, Netherlands
| |
Collapse
|
2
|
Zon M, Ganesh G, Deen MJ, Fang Q. Context-Aware Medical Systems within Healthcare Environments: A Systematic Scoping Review to Identify Subdomains and Significant Medical Contexts. Int J Environ Res Public Health 2023; 20:6399. [PMID: 37510631 PMCID: PMC10379857 DOI: 10.3390/ijerph20146399] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Received: 04/03/2023] [Revised: 06/24/2023] [Accepted: 07/04/2023] [Indexed: 07/30/2023]
Abstract
Context awareness is a field in pervasive computing, which has begun to impact medical systems via an increasing number of healthcare applications that are starting to use context awareness. The present work seeks to determine which contexts are important for medical applications and which domains of context-aware applications exist in healthcare. A systematic scoping review of context-aware medical systems currently used by patients or healthcare providers (inclusion criteria) was conducted between April 2021 and June 2023. A search strategy was designed and applied to Pub Med, EBSCO, IEEE Explore, Wiley, Science Direct, Springer Link, and ACM, articles from the databases were then filtered based on their abstract, and relevant articles were screened using a questionnaire applied to their full texts prior to data extraction. Applications were grouped into context-aware healthcare application domains based on past reviews and screening results. A total of 25 articles passed all screening levels and underwent data extraction. The most common contexts used were user location (8 out of 25 studies), demographic information (6 out of 25 studies), movement status/activity level (7 out of 25 studies), time of day (5 out of 25 studies), phone usage patterns (5 out of 25 studies), lab/vitals (7 out of 25 studies), and patient history data (8 out of 23 studies). Through a systematic review process, the current study determined the key contexts within context-aware healthcare applications that have reached healthcare providers and patients. The present work has illuminated many of the early successful context-aware healthcare applications. Additionally, the primary contexts leveraged by these systems have been identified, allowing future systems to focus on prioritizing the integration of these key contexts.
Collapse
Affiliation(s)
- Michael Zon
- Michael DeGroote School of Medicine, McMaster University, Hamilton, ON L8S 4L8, Canada
- School of Biomedical Engineering, McMaster University, Hamilton, ON L8S 4L8, Canada
| | - Guha Ganesh
- School of Biomedical Engineering, McMaster University, Hamilton, ON L8S 4L8, Canada
| | - M Jamal Deen
- Department of Electrical and Computer Engineering, McMaster University, Hamilton, ON L8S 4L8, Canada
| | - Qiyin Fang
- School of Biomedical Engineering, McMaster University, Hamilton, ON L8S 4L8, Canada
- Department of Engineering Physics, McMaster University, Hamilton, ON L8S 4L8, Canada
| |
Collapse
|