{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,10,10]],"date-time":"2025-10-10T01:43:08Z","timestamp":1760060588351,"version":"build-2065373602"},"reference-count":37,"publisher":"MDPI AG","issue":"17","license":[{"start":{"date-parts":[[2025,9,4]],"date-time":"2025-09-04T00:00:00Z","timestamp":1756944000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Current machine learning approaches focusing on motor symptoms in Parkinson\u2019s disease are commonly trained on small datasets and often lack generalizability from developmental setups to clinical applications. Foundation models using large, unlabeled datasets of healthy participants through self-supervised learning appear attractive for such setups with limited samples, despite the potential impact of motoric symptoms. Acting as an exemplar, this study aims to evaluate the robustness of fine-tuned models in recognizing movements related to motor examinations across datasets and recording setups. Accelerometer data of 51 participants with Parkinson\u2019s disease in three different training and fine-tuning setups were used to tailor the general model to the disease. Training the model on pre-trained weights, both partially (F1 = 0.70) and fully (F1 = 0.69), statistically significantly outperformed training the model from scratch (F1 = 0.55) in a nested cross-validation. For evaluation, the model\u2019s ability to process data recorded from 24 patients in clinic was tested. The models achieved lower mean F1 scores of 0.33 (train from scratch), 0.43 for full, and 0.48 for partial fine-tuning, but demonstrated improved generalizability and robustness regarding the orientation of sensors compared to training from scratch. Utilizing foundation models for accelerometer data trained on healthy participants and fine-tuned for clinical applications in movement disorders appears as an effective strategy for optimized generalizability with small datasets.<\/jats:p>","DOI":"10.3390\/s25175523","type":"journal-article","created":{"date-parts":[[2025,9,5]],"date-time":"2025-09-05T07:46:17Z","timestamp":1757058377000},"page":"5523","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Assessing the Generalizability of Foundation Models for the Recognition of Motor Examinations in Parkinson\u2019s Disease"],"prefix":"10.3390","volume":"25","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-9301-8872","authenticated-orcid":false,"given":"Christopher","family":"Gundler","sequence":"first","affiliation":[{"name":"Institute for Applied Medical Informatics, University Medical Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-1987-9427","authenticated-orcid":false,"given":"Alexander Johannes","family":"Wiederhold","sequence":"additional","affiliation":[{"name":"Institute for Applied Medical Informatics, University Medical Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7680-2147","authenticated-orcid":false,"given":"Monika","family":"P\u00f6tter-Nerger","sequence":"additional","affiliation":[{"name":"Department of Neurology, University Medical Center Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2025,9,4]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"939","DOI":"10.1016\/S1474-4422(18)30295-3","article-title":"Global, Regional, and National Burden of Parkinson\u2019s Disease, 1990\u20132016: A Systematic Analysis for the Global Burden of Disease Study 2016","volume":"17","author":"Dorsey","year":"2018","journal-title":"Lancet Neurol."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"1591","DOI":"10.1002\/mds.26424","article-title":"MDS Clinical Diagnostic Criteria for Parkinson\u2019s Disease","volume":"30","author":"Postuma","year":"2015","journal-title":"Mov. Disord."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"111","DOI":"10.1146\/annurev-bioeng-062117-121036","article-title":"New Sensor and Wearable Technologies to Aid in the Diagnosis and Treatment Monitoring of Parkinson\u2019s Disease","volume":"21","author":"Monje","year":"2019","journal-title":"Annu. Rev. Biomed. Eng."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1371","DOI":"10.1080\/14737175.2021.2000392","article-title":"Adaptive, Personalized Closed-Loop Therapy for Parkinson\u2019s Disease: Biochemical, Neurophysiological, and Wearable Sensing Systems","volume":"21","author":"Tinkhauser","year":"2021","journal-title":"Expert Rev. Neurother."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"5313","DOI":"10.1007\/s00415-023-11873-1","article-title":"Machine Learning for Adaptive Deep Brain Stimulation in Parkinson\u2019s Disease: Closing the Loop","volume":"270","author":"Oliveira","year":"2023","journal-title":"J. Neurol."},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"Giannakopoulou, K.-M., Roussaki, I., and Demestichas, K. (2022). Internet of Things Technologies and Machine Learning Methods for Parkinson\u2019s Disease Diagnosis, Monitoring and Management: A Systematic Review. Sensors, 22.","DOI":"10.3390\/s22051799"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"1272","DOI":"10.1002\/mds.26642","article-title":"Technology in Parkinson Disease: Challenges and Opportunities","volume":"31","author":"Espay","year":"2016","journal-title":"Mov. Disord."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"120541","DOI":"10.1016\/j.eswa.2023.120541","article-title":"Deep Learning and Wearable Sensors for the Diagnosis and Monitoring of Parkinson\u2019s Disease: A Systematic Review","volume":"229","author":"Sigcha","year":"2023","journal-title":"Expert Syst. Appl."},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"105","DOI":"10.1016\/j.parkreldis.2021.02.006","article-title":"Quantitative Mobility Measures Complement the MDS-UPDRS for Characterization of Parkinson\u2019s Disease Heterogeneity","volume":"84","author":"Hill","year":"2021","journal-title":"Park. Relat. Disord."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"186","DOI":"10.1016\/j.gaitpost.2021.10.029","article-title":"Surrogates for Rigidity and PIGD MDS-UPDRS Subscores Using Wearable Sensors","volume":"91","author":"Safarpour","year":"2022","journal-title":"Gait Posture"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"109","DOI":"10.1002\/mds.27830","article-title":"Home-Based Monitoring of Falls Using Wearable Sensors in Parkinson\u2019s Disease","volume":"35","author":"Smits","year":"2020","journal-title":"Mov. Disord."},{"key":"ref_12","first-page":"429","article-title":"Systematic Review Looking at the Use of Technology to Measure Free-Living Symptom and Activity Outcomes in Parkinson\u2019s Disease in the Home or a Home-like Environment","volume":"10","author":"Morgan","year":"2020","journal-title":"J. Park. Dis."},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Tsanas, A., and Triantafyllidis, A. (2023). Up-Sampling Active Learning: An Activity Recognition Method for Parkinson\u2019s Disease Patients. Proceedings of the Pervasive Computing Technologies for Healthcare, Springer Nature.","DOI":"10.1007\/978-3-031-34586-9"},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Cheng, W.-Y., Scotland, A., Lipsmeier, F., Kilchenmann, T., Jin, L., Schjodt-Eriksen, J., Wolf, D., Zhang-Schaerer, Y.-P., Garcia, I.F., and Siebourg-Polster, J. (2017, January 17\u201319). Human Activity Recognition from Sensor-Based Large-Scale Continuous Monitoring of Parkinson\u2019s Disease Patients. Proceedings of the 2017 IEEE\/ACM International Conference on Connected Health: Applications, Systems and Engineering Technologies (CHASE), Philadelphia, PA, USA.","DOI":"10.1109\/CHASE.2017.87"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"pzac129","DOI":"10.1093\/ptj\/pzac129","article-title":"Daily-Living Freezing of Gait as Quantified Using Wearables in People With Parkinson Disease: Comparison with Self-Report and Provocation Tests","volume":"102","author":"Denk","year":"2022","journal-title":"Phys. Ther."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"52","DOI":"10.1186\/s12984-020-00684-4","article-title":"Role of Data Measurement Characteristics in the Accurate Detection of Parkinson\u2019s Disease Symptoms Using Wearable Sensors","volume":"17","author":"Shawen","year":"2020","journal-title":"J. Neuroeng. Rehabil."},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Liang, Y., Wen, H., Nie, Y., Jiang, Y., Jin, M., Song, D., Pan, S., and Wen, Q. (2024, January 25\u201329). Foundation Models for Time Series Analysis: A Tutorial and Survey. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Barcelona, Spain.","DOI":"10.1145\/3637528.3671451"},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"2288","DOI":"10.1109\/JBHI.2022.3144917","article-title":"Estimating Parkinsonism Severity in Natural Gait Videos of Older Adults with Dementia","volume":"26","author":"Sabo","year":"2022","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Guo, Y., Huang, D., Zhang, W., Wang, L., Li, Y., Olmo, G., Wang, Q., Meng, F., and Chan, P. (2022). High-Accuracy Wearable Detection of Freezing of Gait in Parkinson\u2019s Disease Based on Pseudo-Multimodal Features. Comput. Biol. Med., 146.","DOI":"10.1016\/j.compbiomed.2022.105629"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"e26305","DOI":"10.2196\/26305","article-title":"Detecting Parkinson Disease Using a Web-Based Speech Task: Observational Study","volume":"23","author":"Rahman","year":"2021","journal-title":"J. Med. Internet Res."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Zhang, Y., Lei, H., Huang, Z., Li, Z., Liu, C.-M., and Lei, B. (2022, January 21\u201325). Parkinson\u2019s Disease Classification with Self-Supervised Learning and Attention Mechanism. Proceedings of the 2022 26th International Conference on Pattern Recognition (ICPR), Montreal, QC, Canada.","DOI":"10.1109\/ICPR56361.2022.9956213"},{"key":"ref_22","unstructured":"Wang, L., Dou, Q., Fletcher, P.T., Speidel, S., and Li, S. (2022, January 18\u201322). GaitForeMer: Self-Supervised Pre-Training of Transformers via Human Motion Forecasting for Few-Shot Gait Impairment Severity Estimation. Proceedings of the Medical Image Computing and Computer Assisted Intervention\u2014MICCAI 2022, Resorts World Sentosa, Singapore."},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Jiang, H., Bryan Lim, W.Y., Shyuan Ng, J., Wang, Y., Chi, Y., and Miao, C. (2021, January 6\u201311). Towards Parkinson\u2019s Disease Prognosis Using Self-Supervised Learning and Anomaly Detection. Proceedings of the ICASSP 2021\u20142021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Toronto, ON, Canada.","DOI":"10.1109\/ICASSP39728.2021.9414840"},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"S\u00e1nchez-Fern\u00e1ndez, L.P., Garza-Rodr\u00edguez, A., S\u00e1nchez-P\u00e9rez, L.A., and Mart\u00ednez-Hern\u00e1ndez, J.M. (2023). A Computer Method for Pronation-Supination Assessment in Parkinson\u2019s Disease Based on Latent Space Representations of Biomechanical Indicators. Bioengineering, 10.","DOI":"10.3390\/bioengineering10050588"},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Xia, Y., Sun, H., Zhang, B., Xu, Y., and Ye, Q. (2024). Prediction of Freezing of Gait Based on Self-Supervised Pretraining via Contrastive Learning. Biomed. Signal Process. Control., 89.","DOI":"10.1016\/j.bspc.2023.105765"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"91","DOI":"10.1038\/s41746-024-01062-3","article-title":"Self-Supervised Learning for Human Activity Recognition Using 700,000 Person-Days of Wearable Data","volume":"7","author":"Yuan","year":"2023","journal-title":"npj Digit. Med."},{"key":"ref_27","unstructured":"Zhang, Y., Ayush, K., Qiao, S., Heydari, A.A., Narayanswamy, G., Xu, M.A., Metwally, A.A., Xu, S., Garrison, J., and Xu, X. (2025). SensorLM: Learning the Language of Wearable Sensors. arXiv."},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"95","DOI":"10.1038\/s41746-019-0169-y","article-title":"Feasibility and Utility of a Clinician Dashboard from Wearable and Mobile Application Parkinson\u2019s Disease Data","volume":"2","author":"Elm","year":"2019","journal-title":"NPJ Digit. Med."},{"key":"ref_29","unstructured":"(2025, August 31). Synapse.org. MJFF Levodopa Response Study. Available online: https:\/\/www.synapse.org\/Synapse:syn20681023\/wiki\/594678."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"48","DOI":"10.1038\/s41597-021-00830-0","article-title":"Accelerometer Data Collected with a Minimum Set of Wearable Sensors from Subjects with Parkinson\u2019s Disease","volume":"8","author":"Daneault","year":"2021","journal-title":"Sci. Data"},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"47","DOI":"10.1038\/s41597-021-00830-0","article-title":"Limb and Trunk Accelerometer Data Collected with Wearable Sensors from Subjects with Parkinson\u2019s Disease","volume":"8","author":"Daneault","year":"2021","journal-title":"Sci. Data"},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"53","DOI":"10.1038\/s41746-021-00414-7","article-title":"Crowdsourcing Digital Health Measures to Predict Parkinson\u2019s Disease Severity: The Parkinson\u2019s Disease Digital Biomarker DREAM Challenge","volume":"4","author":"Sieberts","year":"2021","journal-title":"NPJ Digit. Med."},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Wiederhold, A.J., Zhu, Q.R., Spiegel, S., Dadkhah, A., P\u00f6tter-Nerger, M., Langebrake, C., \u00dcckert, F., and Gundler, C. (2025). Opportunities and Limitations of Wrist-Worn Devices for Dyskinesia Detection in Parkinson\u2019s Disease. Sensors, 25.","DOI":"10.3390\/s25144514"},{"key":"ref_34","unstructured":"(2025, August 26). Activeinsights. GENEActiv with Software: Instructions for Use. Available online: https:\/\/activinsights.com\/wp-content\/uploads\/2024\/09\/GENEActiv-1.2-IFU-rev-6.pdf."},{"key":"ref_35","first-page":"22","article-title":"A Unified Data Architecture for Assessing Motor Symptoms in Parkinson\u2019s Disease","volume":"307","author":"Gundler","year":"2023","journal-title":"Stud. Health Technol. Inform."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"304","DOI":"10.1109\/TIFS.2018.2850032","article-title":"Evaluation of Algorithms for Orientation Invariant Inertial Gait Matching","volume":"14","author":"Subramanian","year":"2019","journal-title":"IEEE Trans. Inf. Forensics Secur."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3563948","article-title":"Transfer Learning for Human Activity Recognition Using Representational Analysis of Neural Networks","volume":"4","author":"An","year":"2023","journal-title":"ACM Trans. Comput. Healthc."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/25\/17\/5523\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,9]],"date-time":"2025-10-09T18:39:47Z","timestamp":1760035187000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/25\/17\/5523"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,4]]},"references-count":37,"journal-issue":{"issue":"17","published-online":{"date-parts":[[2025,9]]}},"alternative-id":["s25175523"],"URL":"https:\/\/doi.org\/10.3390\/s25175523","relation":{},"ISSN":["1424-8220"],"issn-type":[{"type":"electronic","value":"1424-8220"}],"subject":[],"published":{"date-parts":[[2025,9,4]]}}}