Abstract
Background
Patients with essential tremor (ET), Parkinson’s disease (PD) and dystonic tremor (DT) can be difficult to classify and often share similar characteristics.
Objectives
To use ubiquitous smartphone accelerometers with and without clinical features to automate tremor classification using supervised machine learning, and to use unsupervised learning to evaluate if natural clusterings of patients correspond to assigned clinical diagnoses.
Methods
A supervised machine learning classifier was trained to classify 78 tremor patients using leave-one-out cross-validation to estimate performance on unseen accelerometer data. An independent cohort of 27 patients were also studied. Next, we focused on a subset of 48 patients with both smartphone-based tremor measurements and detailed clinical assessment metrics and compared two separate machine learning classifiers trained on these data.
Results
The classifier yielded a total accuracy of 74.4% and F1-score of 0.74 for a trinary classification with an area under the curve of 0.904, average F1-score of 0.94, specificity of 97% and sensitivity of 84% in classifying PD from ET or DT. The algorithm classified ET from non-ET with 88% accuracy, but only classified DT from non-DT with 29% accuracy. A poorer performance was found in the independent cohort. Classifiers trained on accelerometer and clinical data respectively obtained similar results.
Conclusions
Machine learning classifiers achieved a high accuracy of PD, however moderate accuracy of ET, and poor accuracy of DT classification. This underscores the difficulty of using AI to classify some tremors due to lack of specificity in clinical and neuropathological features, reinforcing that they may represent overlapping syndromes.
Similar content being viewed by others
References
Martino D, Espay AJ, Fasano A, Morgante F (2016) Rhythmical involuntary movements (tremor and tremor-like conditions). In: Martino D, Espay AJ, Fasano A, Morgante F (eds) Disorders of movement—a guide to diagnosis and treatment. Springer, Berlin, pp 207–263
Ure RJ, Dhanju S, Lang AE, Fasano A (2016) Unusual tremor syndromes: know in order to recognise. J Neurol Neurosurg Psychiatry 87(11):1191–1203. https://doi.org/10.1136/jnnp-2015-311693 (Epub 2016 Mar 16)
Rizzo G, Copetti M, Arcuti S, Martino D, Fontana A, Logroscino G (2016) Accuracy of clinical diagnosis of Parkinson disease A systematic review and meta-analysis. Neurology 86:566–576
Bhatia KP, Bain P, Bajaj N et al (2018) Consensus Statement on the classification of tremors. from the task force on tremor of the International Parkinson and Movement Disorder Society: IPMDS Task Force on Tremor Consensus Statement. Mov Disord 33(1):75–87. https://doi.org/10.1002/mds.27121
Fasano A, Lang AE, Espay AJ (2018) What is “Essential” about essential tremor? A diagnostic placeholder. Mov Disord 33(1):58–61. https://doi.org/10.1002/mds.27288
Rajalingam R, Breen DP, Lang AE, Fasano A (2018) Essential tremor plus is more common than essential tremor: Insights from the reclassification of a cohort of patients with lower limb tremor. Parkinsonism Relat Disord 56:109–110. https://doi.org/10.1016/j.parkreldis.2018.06.029
Fearon C, Espay AJ, Lang AE et al (2019) Soft signs in movement disorders: friends or foes? J Neurol Neurosurg Psychiatry 90(8):961–962. https://doi.org/10.1136/jnnp-2018-318455
Louis ED (2016) Diagnosis and management of tremor. Continuum (Minneap Minn) 22:1143–1158
Bove F, Di Lazzaro G, Mulas D, Cocciolillo F, Di Giuda D, Bentivoglio AR (2018) A role for accelerometry in the differential diagnosis of tremor syndromes. Funct Neurol 33(1):45–49
Mourcou Q, Fleury A, Franco C, Klopcic F, Vuillerme N (2015) Performance evaluation of smartphone inertial sensors measurement for range of motion. Sensors 15(9):23168–23187. https://doi.org/10.3390/s150923168
Balachandar A, Fasano A (2017) Characterizing orthostatic tremor using a smartphone application. Tremor Other Hyperkinet Mov 7:488. https://doi.org/10.7916/D8V12GRJ.eCollection2017
Bhatti D, Thompson R, Hellman A, Penke C, Bertoni JM, Torres-Russotto D (2017) smartphone apps provide a simple, accurate bedside screening tool for orthostatic tremor. Move Disord Clin Pract 4(6):852–857. https://doi.org/10.1002/mdc3.12547
López-Blanco R, Velasco MA, Méndez-Guerrero A et al (2019) Smartwatch for the analysis of rest tremor in patients with Parkinsons disease. J Neurol Sci 401:37–42. https://doi.org/10.1016/j.jns.2019.04.011
di Biase L, Brittain JS, Shah SA et al (2017) Tremor stability index: a new tool for differential diagnosis in tremor syndromes. Brain 140:1977–1986
Daneault JF, Carignan B, Codere CE, Sadikot AF, Duval C (2013) Using a smart phone as a standalone platform for detection and monitoring of pathological tremors. Front Hum Neurosci. https://doi.org/10.3389/fnhum.2012.00357
Barrantes S, Sánchez Egea AJ, González Rojas HA et al (2017) Differential diagnosis between Parkinson’s disease and essential tremor using the smartphone’s accelerometer. PLoS One 12:e0183843
Jones E, Oliphant E, Peterson P (2001) SciPy: Open source scientific tools for python. http://www.scipy.org/
Oliphant TE (2007) Python for scientific computing. Comput Sci Eng 9(3):10–20. https://doi.org/10.1109/mcse.2007.58
Pedregosa F, Varoquaux G, Gramfort A et al (2011) Scikit-learn: machine learning in python. JMLR 12:2825–2830
Fahn S, Tolosa E, Conceppcion M (1993) Clinical rating scale for tremor. In: Jankovic J, Tolosa E (eds) Parkinson’s disease and movement disorders. Williams and Wilkins, Baltimore, pp 271–280
Goetz CG, Tilley BC, Shaftman SR et al (2008) Movement Disorder Society-sponsored revision of the Unified Parkinsons Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov Disord 23(15):2129–2170. https://doi.org/10.1002/mds.22340
Burke RE, Fahn S, Marsden CD, Bressman SB, Moskowitz C, Friedman J (1985) Validity and reliability of a rating scale for the primary torsion dystonias. Neurology 35:73–77
Schmitz-Hübsch T, du Montcel TS, Baliko L et al (2006) Scale for the assessment and rating of ataxia. Neurology 66:1717–1720. https://doi.org/10.1212/01.wnl.0000219042.60538.92
Espay AJ, Li JY, Johnston L, Chen R, Lang AE (2005) Mirror movements in parkinsonism: evaluation of a new clinical sign. J Neurol Neurosurg Psychiatry 76:1355–1358
Abdo WF, Borm GF, Munneke M, Verbeek MM, Esselink RAJ, Bloem BR (2006) Ten steps to identify atypical parkinsonism. J Neurol Neurosurg Psychiatry 77(12):1367–1369. https://doi.org/10.1136/jnnp.2006.091322
Sternberg EJ, Alcalay RN, Levy OA, Louis ED. The “head snap”: a subtle clinical feature during the finger-nose-finger maneuver in essential tremor. Tremor Other Hyperkinet Mov (N Y). 2013; 3: pii: tre-03–159–3719–1
Kulkarni O, Lafaver K, Tarsy D (2013) The, “floating door sign” in Parkinson’s disease. Parkinsonism Relat Disord 19:825–826
Lieberman A, Deep A, Shi J et al (2017) Downward finger displacement distinguishes Parkinson disease dementia from Alzheimer disease. Int J Neurosci 128(2):151–154. https://doi.org/10.1080/00207454.2017.1379518
Bajaj NP, Knöbel M, Gontu V, Bain PG (2011) Can spiral analysis predict the FP-CIT SPECT scan result in tremulous patients? Mov Disord 26(4):699–704. https://doi.org/10.1002/mds.23507
Krismer F, Pinter B, Mueller C et al (2017) Sniffing the diagnosis: olfactory testing in neurodegenerative parkinsonism. Parkinsonism Relat Disord 35:36–41. https://doi.org/10.1016/j.parkreldis.2016.11.010
Stiasny-Kolster K, Mayer G, Schafer S, Moller JC, Heinzel-Gutenbrunner M, Oertel WH (2007) The REM sleep behavior disorder screening questionnaire—a new diagnostic instrument. Mov Disord 22:2386–2393
Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J (1961) An inventory for measuring depression. Arch Gen Psychiatry 4:561–571
Nasreddine ZS, Phillips NA, Bédirian V et al (2005) The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53(4):695–699. https://doi.org/10.1111/j.1532-5415.2005.53221.x
Visser M, Marinus J, Stiggelbout AM, Hilten JJV (2004) Assessment of autonomic dysfunction in Parkinsons disease: the SCOPA-AUT. Mov Disord 19(11):1306–1312. https://doi.org/10.1002/mds.20153
McInnes L, Healy J (2018) UMAP: uniform manifold approximation and projection for dimension reduction. ArXiv e-prints. https://arxiv.org/1802.0342
Kearns M, Ron D (1999) Algorithmic stability and sanity-check bounds for leave-one-out cross-validation. Neural Comput 11:1427–1453
Powers DM (2011) Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. J Mach Learn Technol 2(1):37–63
Lipton ZC, Elkan C, Naryanaswamy B (2014) Optimal thresholding of classifiers to maximize F1 measure. In: Machine learning and knowledge discovery in databases lecture notes in computer science. pp 225–239. doi: https://doi.org/10.1007/978-3-662-44851-9_15
Sokolova M, Lapalme G (2009) A systematic analysis of performance measures for classification tasks. Inf Process Manage 45:427–437
Saito T, Rehmsmeier M (2015) The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS ONE. https://doi.org/10.1371/journal.pone.0118432
Cohen JP, Cao T, Viviano JD, Huang CW et al (2021) Problems in the deployment of machine-learned models in health care. CMAJ 193(35):E1391–E1394. https://doi.org/10.1503/cmaj.202066 (Epub 2021 Aug 30)
Knoebel M, Bain PG (2009) Spirography in essential tremor and parkinsonism. Mov Disord 249(24 Suppl. 1):S503
Shah M, Muhammed N, Findley LJ, Hawkes CH (2008) Olfactory tests in the diagnosis of essential tremor. Parkinsonism Relat Disord 14:563–568
Schneider SA, Edwards MJ, Mir P et al (2007) Patients with adult-onset dystonic tremor resembling parkinsonian tremor have scans without evidence of dopaminergic deficit (SWEDDs). Mov Disord 22:2210–2215
Elble RJ (2013) Defining dystonic tremor. Curr Neuropharmacol 11:48–52
Figueroa RL, Zeng-Treitler Q, Kandula S, Ngo LH (2012) Predicting sample size required for classification performance. BMC Med Inform Decis Making. https://doi.org/10.1186/1472-6947-12-8
Sun C, Shrivastava A, Singh S, Gupta A (2017) Revisiting unreasonable effectiveness of data in deep learning era. ArXiv e-prints. http://arxiv.org/1707.02968
He H, Garcia EA (2009) Learning from imbalanced data. IEEE Trans Knowl Data Eng. https://doi.org/10.1109/TKDE.2008.239
Branco P, Torgo L, Ribeiro RP (2016) A survey of predictive modeling on imbalanced domains. ACM Comput Surv 49(2):50. https://doi.org/10.1145/2907070
Fasano A, Mancini M (2020) How Complex is Reality? Wearable-based mobility monitoring: the long road ahead. Lancet Neurol. https://doi.org/10.1016/S1474-4422(20)30033-8 (Epub ahead of print)
Elble RJ, McNames J (2016) Using portable transducers to measure tremor severity. Tremor Other Hyperkinet Mov (N Y) 6:375
López-Blanco R, Velasco MA, Méndez-Guerrero A et al (2018) Essential tremor quantification based on the combined use of a smartphone and a smartwatch: the NetMD study. J Neurosci Methods 303:95–102. https://doi.org/10.1016/j.jneumeth.2018.02.015
Acknowledgements
Authors are grateful to Dr. Hamza Jalal for the help in setting up the experimental protocol.
Funding
AB, MA, LM, AS, WDH, AB—Stock Ownership in medically-related fields: none, Intellectual Property Rights: none, Consultancies: none, Expert Testimony: none, Advisory Boards: none, Employment: none, Partnerships: none, Contracts: none, Honoraria: none, Royalties: none, Grants: none, Other: none. LO—Stock Ownership in medically-related fields: none, Intellectual Property Rights: none, Consultancies: none, Expert Testimony: none, Advisory Boards: none, Employment: none, Partnerships: none, Contracts: none, Honoraria: none, Royalties: none, Grants: none, Other: Funding for travel from Medtronic. AF—Stock Ownership in medically-related fields: none, Intellectual Property Rights: none, Consultancies: Abbvie, Medtronic, Boston Scientific, Sunovion, Chiesi farmaceutici, UCB, Ipsen, Expert Testimony: none, Advisory Boards: Abbvie, Boston Scientific, Ipsen, Employment: none, Partnerships: none, Contracts: none, Honoraria: Abbvie, Medtronic, Boston Scientific, Sunovion, Chiesi farmaceutici, UCB, Ipsen, Royalties: none, Grants: University of Toronto, Weston foundation, Abbvie, Medtronic, Boston Scientific, Other: none.
Author information
Authors and Affiliations
Contributions
AB—Research project: execution; Statistical analysis: design, execution; Manuscript: writing of the first draft. MA—Research project: execution; Manuscript: review and critique. LO—Research project: execution; Manuscript: review and critique. LM—Research project: execution; Manuscript: review and critique. AM— Research project: execution; Manuscript: review and critique. AS—Research project: organization, execution; Statistical analysis: review and critique; Manuscript: review and critique. AJE—Research project: organization, execution; Statistical analysis: review and critique; Manuscript: review and critique. WDH—Statistical analysis: review and critique; Manuscript: review and critique. AB—Statistical analysis: execution, review and critique; Manuscript: review and critique. FR—Research project: conception; Statistical analysis: design, review and critique; Manuscript: review and critique. AF—Research project: conception, organization; Statistical analysis: review and critique; Manuscript: review and critique.
Corresponding author
Ethics declarations
Conflicts of interest
AJE is chair of the Task Force on Technology of the IPMDS. He has received research support from Great Lakes Neurotechnology. AF sits in the advisory board of Evotion and received honoraria from Apple for an unrelated project.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Balachandar, A., Algarni, M., Oliveira, L. et al. Are smartphones and machine learning enough to diagnose tremor?. J Neurol 269, 6104–6115 (2022). https://doi.org/10.1007/s00415-022-11293-7
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00415-022-11293-7