Objective: The aim of the study is to determine whether the risk of falling can be detected in the gait of a patient with PD using the OptoGait system and machine learning (ML) algorithms.

Background: The risk of falling normally increases as PD progresses, posing a significant risk of accidents and hospital care [1] [2] [3]. With the OptoGait gait analysis device, gait can be tracked using optical sensors to measure gait parameters [7]. The device produces processed gait data and ML can be used to analyse the data to detect patterns that would otherwise be difficult to discern. An example of this is the risk of falling in Parkinson’s patients [8].

Method: The data was collected from patients with PD in 2023-2024.

✓ Group 1: people with PD for more than 4 years with a tendency to fall (at least 2 falls in the past year)

✓ Group 2: people with PD without a tendency to fall

✓ Control group: people over 18 years of age without PD and without a tendency to fall

The walk (20m) was recorded using the OptoGait. The following data were collected from patient records: H&Y, gender, age, height, weight and shoe size. Analysis was made with classification algorithms [4] with Python.

Results: Algorithms were trained using training data to classify individual steps of walkers into predefined categories: fallers, non-fallers, and control subjects [figure1]. The performance of the training was evaluated using new data with the k-fold cross-validation method. The best-performing algorithms were based on decision trees, such as Random Forest (acc. 92,0%) [figure2]. The results varied significantly across the different algorithms – Decision Tree 91%, k-NN 75%, SVM 61%, Kernel SVM 66%, Naïve Bayes 55%, Logistic regression 61%. The most important parameters were stance phase, preswing and imbalance.

Conclusion: Group 1 accounts for the highest classification errors, with the model correctly identifying only 71% of cases. This may be due to the challenge of distinguishing Group 1 from Group 2, as some fallers exhibit greater gait instability than others. Additionally, some control subjects demonstrate increased unsteadiness due to age, leading to misclassification as non-falling PD patients. PD also typically presents with asymmetric symptoms, which should also be reflected in gait patterns [5] [6]. However, step asymmetry is not well captured in the data, necessitating further analysis.

Figure 1

Figure 2

References: [1] Cao P, Min CH. Fall Prediction in People with Parkinson’s Disease. Annu Int Conf IEEE Eng Med Biol Soc. 2022 Jul;2022:1502-1505. doi: 10.1109/EMBC48229.2022.9872013. PMID: 36085756.
[2] Ginis P, Nackaerts E, Nieuwboer A, Heremans E. Cueing for people with Parkinson’s disease with freezing of gait: A narrative review of the state-of-the-art and novel perspectives. Ann Phys Rehabil Med. 2018 Nov;61(6):407-413. doi: 10.1016/j.rehab.2017.08.002. Epub 2017 Sep 7. PMID: 28890341.
[3] Hausdorff, J. & Buchman, A. (2013) What Links Gait Speed and MCI With Dementia? A Fresh Look at the Association Between Motor and Cognitive Function. J Gerontol A Biol Sci Med Sci. 2013 Apr; 68(4): 409–411. DOI: 10.1093/gerona/glt002
[4] Juutinen, M., Wang, C., Zhu, J., Haladjian, J., Ruokolainen, J., Puustinen, J. & Vehkaoja, A. (2020) Parkinson’s disease detection from 20-step walking tests using inertial sensors of a smartphone : machine learning approach based on an observational case-control study. PLoS ONE, 2020:7. DOI: 10.1371/journal.pone.0236258
[5] Krebs DE, Edelstein JE, Fishman S. Reliability of observational kinematic gait analysis. Phys Ther. 1985 Jul;65(7):1027-33. doi: 10.1093/ptj/65.7.1027.
[6] Patla AE, Proctor J, Morson B 1987 Observations on aspects of visual gait assessment: a questionnaire study. Physiotherapy Canada 39(5): 311-316
[7] Microgate Corporation. N.d. ‘What is OptoGait?’ Website. Cited 18.6.2023. http://optogait.com/What-is-OptoGait.
[8] Morley, J. (2016) Gait, Falls, and Dementia. Journal of the American Medical Directors Association. Volume 17, Issue 6, 1 June 2016,
Pages 467-470. DOI: 10.1016/j.jamda.2016.03.024

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/analysis-of-optogait-walking-data-using-machine-learning-models-using-classification-algorithms-to-assess-the-risk-of-falling-in-parkinsons-patients/