Quantitative Assessment of Ataxia in Multiple Sclerosis Patients using Spatiotemporal Parameters: A Relief-Based Machine Learning Analysis

Abstract

Background and Objective Multiple sclerosis (MS) is a chronic, progressive, and autoimmune disease of the central nervous system (CNS) characterized by inflammation, demyelination, and axonal injury. In patients with newly diagnosed MS (ndMS), ataxia can present either as mild or severe and can be difficult to diagnose in the absence of clinical disability. Such difficulties can be eliminated by using decision support systems supported by machine learning methods. The present study aimed to achieve early diagnosis of ataxia in ndMS patients by using machine learning methods with spatiotemporal parameters.

Materials and Methods The prospective study included 32 ndMS patients with an Expanded Disability Status Scale (EDSS) score of≤2.0 and 32 healthy volunteers. A total of 14 parameters were elicited by using a Win-Track platform. The ndMS patients were differentiated from healthy individuals using multiple classifiers including Artificial Neural Network (ANN), Support Vector Machine (SVM), the k-nearest neighbors (K-NN) algorithm, and Decision Tree Learning (DTL). To improve the performance of the classification, a Relief-based feature selection algorithm was applied to select the subset that best represented the whole dataset. Performance evaluation was achieved based on several criteria such as Accuracy (ACC), Sensitivity (SN), Specificity (SP), and Precision (PREC).

Results ANN had a higher classification performance compared to other classifiers, whereby it provided an accuracy, sensitivity, and specificity of 89, 87.8, 90.3% with the use of all parameters and provided the values of 93.7, 96.6%, and 91.1% with the use of parameters selected by the Relief algorithm, respectively.

Significance To our knowledge, this is the first study of its kind in the literature to investigate the diagnosis of ataxia in ndMS patients by using machine learning methods with spatiotemporal parameters. The proposed method, i. e. Relief-based ANN method, successfully diagnosed ataxia by using a lower number of parameters compared to the numbers of parameters reported in clinical studies, thereby reducing the costs and increasing the performance of the diagnosis. The method also provided higher rates of accuracy, sensitivity, and specificity in the diagnosis of ataxia in ndMS patients compared to other methods. Taken together, these findings indicate that the proposed method could be helpful in the diagnosis of ataxia in minimally impaired ndMS patients and could be a pathfinder for future studies.

Zusammenfassung

Hintergrund und Ziel Multiple Sklerose (MS) ist eine chronische, fortschreitende und Autoimmunerkrankung des Zentralnervensystems (ZNS), die durch Entzündung, Demyelinisierung und axonale Verletzung gekennzeichnet ist. Bei Patienten mit neu diagnostizierter MS (ndMS) kann Ataxie entweder leicht oder schwer auftreten und ohne klinische Behinderung schwierig zu diagnostizieren sein. Solche Schwierigkeiten können durch die Verwendung von Entscheidungsunterstützungssystemen beseitigt werden, die durch Methoden des maschinellen Lernens unterstützt werden. Die vorliegende Studie zielte darauf ab, eine frühzeitige Diagnose der Ataxie zu erreichen bei ndMS-Patienten mithilfe von Methoden des maschinellen Lernens mit raumzeitlichen Parametern zu bewerten.

Material und Methoden Die prospektive Studie umfasste 32 ndMS-Patienten mit einem EDSS-Wert (Expanded Disability Status Scale) von≤2,0 und 32 gesunde Freiwillige. Insgesamt 14 Parameter wurden mithilfe einer Win-Track-Plattform ermittelt. Die ndMS-Patienten wurden von gesunden Personen unter Verwendung mehrerer Klassifikatoren unterschieden, einschließlich des künstlichen neuronalen Netzwerks (KNN), der Support Vector Machine (SVM), des Algorithmus für k-nächste Nachbarn (K-NN) und des Decision Tree Learning (DTL). Um die Leistung der Klassifizierung zu verbessern, wurde ein auf Relief basierender Algorithmus zur Merkmalsauswahl angewendet, um die Teilmenge auszuwählen, die den gesamten Datensatz am besten repräsentiert. Die Leistungsbewertung wurde anhand verschiedener Kriterien wie Regelgenauigkeit, Empfindlichkeit, Spezifität und Präzision erreicht.

Ergebnisse ANN hatte im Vergleich zu anderen Klassifikatoren eine höhere Klassifizierungsleistung, wobei es bei Verwendung aller Parameter eine Genauigkeit, Sensitivität und Spezifität von 89, 87,8, 90,3% und Werte von 93,7, 96,6% und 91,1 ergab% unter Verwendung von Parametern, die vom Relief-Algorithmus ausgewählt wurden.

Schlussfolgerun Nach unserem Kenntnisstand ist dies die erste Studie dieser Art in der Literatur, die die Diagnose von Ataxie bei ndMS-Patienten mithilfe maschineller Lernmethoden mit raumzeitlichen Parametern untersucht. Das vorgeschlagene Verfahren, d. H. Das entlastungsbasierte ANN-Verfahren, diagnostizierte erfolgreich Ataxie unter Verwendung einer geringeren Anzahl von Parametern im Vergleich zu der Anzahl von Parametern, die in klinischen Studien angegeben wurden, wodurch die Kosten gesenkt und die

Key words

Multiple sclerosis - Ataxia - Machine learning - Relief method - Artificial neural networks

Schlüsselwörter

Multiple Sklerose - Entlastungsmethode - Künstliche Neuronale Netze - Ataxie - maschinelles Lernen

Publication History

Received: 28 December 2020

Accepted: 19 May 2021

Article published online:
23 August 2021

• References

• 1 McDonald WI, Compston A, Edan G. et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol 2001; 50: 121-127 doi:10.1002/ana.1032
• 2 Koch M. What’s going wrong in ataxia and tremor in MS. J MS in Focus 2009; 200: 13
• 3 Eissa A, Lebel RM, Korzan JR. et al. Detecting lesions in multiple sclerosis at 4.7 tesla using phase susceptibility-weighting and T2-weighting. J Magn Reson Imaging 2009; 30: 737-742 doi:10.1002/jmri.21926
• 4 Winser SJ, Hale L, Claydon LS. et al. Outcome measures for the assessment of balance and posture control in cerebellar ataxia. Physical Therapy Reviews 2013; 18: 117-133 doi:10.1179/1743288X13Y.0000000065
• 5 Cassidy E, Kilbride C, Holland A. et al. Management of the Ataxias: towards best Clinical Practice. 2009; https://duncanfoundation.org/wp-content/uploads/2020/03/Physiotherapy_Supplement_to_Ataxia_Guidelines_2010.pdf
• 6 Martin CL, Tan D, Bragge P. et al. Effectiveness of physiotherapy for adults with cerebellar dysfunction: a systematic review. Clin Rehabil 2009; 23: 15-26 doi:10.1177/0269215508097853
• 7 Jacobs JV, Horak FB, Tran VK. et al. Multiple balance tests improve the assessment of postural stability in subjects with Parkinson’s disease. J Neurol Neurosurg Psychiatry 2006; 77: 322-326 doi:10.1136/jnnp.2005.068742
• 8 Berg KO, Wood-Dauphinee SL, Williams JI. et al. Measuring balance in the elderly: preliminary development of an instrument. Physiotherapy Canada 1989; 41: 304-311 doi:10.3138/ptc.41.6.304
• 9 Khasnis A, Gokula RM. Romberg’s test. J Postgrad Med 2003; 49: 169-172 PMID: 12867698
• 10 Martin CL, Phillips BA, Kilpatrick TJ. et al. Gait and balance impairment in early multiple sclerosis in the absence of clinical disability. Mult Scler 2006; 12: 620-628 doi:10.1177/1352458506070658
• 11 Givon U, Zeilig G, Achiron A. Gait analysis in multiple sclerosis: characterization of temporal–spatial parameters using GAITRite functional ambulation system. Gait Posture 2009; 29: 138-142 doi:10.1016/j.gaitpost.2008.07.011
• 12 Heesen C, Böhm J, Reich C. et al. Patient perception of bodily functions in multiple sclerosis: gait and visual function are the most valuable. Mult Scler 2008; 14: 988-991 doi:10.1177/1352458508088916
• 13 Benedetti MG, Piperno R, Simoncini L. et al. Gait abnormalities in minimally impaired multiple sclerosis patients. Mult Scler 1999; 5: 363-368 doi:10.1177/135245859900500510
• 14 Morel E, Allali G, Laidet M. et al. Gait profile score in multiple sclerosis patients with low disability. Gait Posture 2017; 51: 169-173 doi:10.1016/j.gaitpost.2016.10.013
• 15 Kalron A, Dvir Z. Achiron A. Walking while talking—difficulties incurred during the initial stages of multiple sclerosis disease process. Gait Posture 2010; 32: 332-335 doi:10.1016/j.gaitpost.2010.06.002
• 16 DeLisa JA. Gait analysis in the science of rehabilitation. Vol. 2. 1998: Diane Publishing
• 17 Mummolo C, Mangialardi L, Kim JH. Quantifying dynamic characteristics of human walking for comprehensive gait cycle. J Biomech Eng 2013; 135: 91006 doi:10.1115/1.4024755
• 18 Stöckel T, Jacksteit R, Behrens M. et al. The mental representation of the human gait in young and older adults. Front Psychol 2015; 6: 943 doi:10.3389/fpsyg.2015.00943
• 19 Ramachandra P, Maiya AG, Kumar P. Test-Retest Reliability of the Win-Track Platform in Analyzing the Gait Parameters and Plantar Pressures During Barefoot Walking in Healthy Adults. Foot Ankle Spec 2012; 5: 306-312 doi:10.1177/1938640012457680
• 20 Socie MJ, Motl RW, Sosnoff JJ. Examination of spatiotemporal gait parameters during the 6-min walk in individuals with multiple sclerosis. Int J Rehabil Res 2014; 37: 311-316 doi:10.1097/MRR.0000000000000074
• 21 Forman G. An extensive empirical study of feature selection metrics for text classification. Journal of Machine Learning Research 2003; 3 p 1289-1305
• 22 Urbanowicz RJ, Meeker M, La Cava W. et al. Relief-based feature selection: Introduction and review. J Biomed Inform 2018; 85: 189-203 doi:10.1016/j.jbi.2018.07.014
• 23 Kira K, Rendell LA. The feature selection problem: Traditional methods and a new algorithm. in Aaai 1992; 2: 129-134
• 24 He C, Ma M, Wang P. Extract Interpretability-Accuracy balanced Rules from Artificial Neural Networks: A Review. Neurocomputing 2020; 387: 346-358 doi:10.1016/j.neucom.2020.01.036
• 25 Xu M, Papageorgiou DP, Abidi SZ. et al. A deep convolutional neural network for classification of red blood cells in sickle cell anemia. PLoS Comput Biol 2017; 13: e1005746.2017 doi:10.1371/journal.pcbi.1005746
• 26 Luque A, Gómez-Bellido J, Carrasco A. et al. Optimal Representation of Anuran Call Spectrum in Environmental Monitoring Systems Using Wireless Sensor Networks. Sensors (Basel) 2018; 18: 1803 doi:10.3390/s18061803
• 27 Gehlsen G, Beekman K, Assmann N. et al. Gait characteristics in multiple sclerosis: progressive changes and effects of exercise on parameters. Arch Phys Med Rehabil 1986; 67: 536-539 doi:10.5555/uri:pii:0003999386905496
• 28 Noseworthy JH, Vandervoort MK, Wong CJ. et al. Interrater variability with the Expanded Disability Status Scale (EDSS) and Functional Systems (FS) in a multiple sclerosis clinical trial. The Canadian Cooperation MS Study Group. Neurology 1990; 40: 971-975 doi:10.1212/WNL.40.6.971
• 29 Petrov AM, Stolyarov ID, Shkilnyuk GG. et al. Time course of changes in the development of gait disorders in multiple sclerosis. J Neurology, Neuropsychiatry, Psychosomatics 2015; 7: 27-32 doi:10.14412/2074-2711-2015-1-27-32
• 30 Meyer PF, Oddsson LI, De Luca CJ. The role of plantar cutaneous sensation in unperturbed stance. Exp Brain Res 2004; 156: 505-512 doi:10.1007/s00221-003-1804-y
• 31 Palliyath S, Hallett M, Thomas SL. et al. Gait in patients with cerebellar ataxia. Mov Disord 1998; 13: 958-964 doi:10.1002/mds.870130616
• 32 Pau M, Coghe G, Corona F. et al. Effect of spasticity on kinematics of gait and muscular activation in people with multiple sclerosis. J Neurol Sci 2015; 358: 339-344 doi:10.1016/j.jns.2015.09.352
• 33 Buckley E, Mazzà C, McNeill A. A systematic review of the gait characteristics associated with Cerebellar Ataxia. Gait Posture 2018; 60: 154-163 doi:10.1016/j.gaitpost.2017.11.024
• 34 Bethoux F, Bennett S. Bennett, Evaluating walking in patients with multiple sclerosis: which assessment tools are useful in clinical practice?. Int J MS Care 2011; 13: 4-14 doi:10.7224/1537-2073-13.1.4
• 35 Orsnes GB, Sørensen PS, Larsen TK. et al. Effect of baclofen on gait in spastic MS patients. Acta Neurol Scand 2000; 101: 244-248 doi:10.1034/j.1600-0404.2000.101004244x./
• 36 Morton SM, Bastian AJ. Relative contributions of balance and voluntary leg-coordination deficits to cerebellar gait ataxia. J Neurophysiol 2003; 89: 1844-1856 doi:10.1152/jn.00787.2002
• 37 Morris ME, Cantwell C, Vowels L. et al. Changes in gait and fatigue from morning to afternoon in people with multiple sclerosis. J Neurol Neurosurg Psychiatry 2002; 72: 361-365 doi:10.1136/jnnp.72.3.361
• 38 Rodgers MM, Mulcare JA, King DL. et al. Gait characteristics of individuals with multiple sclerosis before and after a 6-month aerobic training program. J Rehabil Res Dev 1999; 36: 183-188 PMID: 10659801
• 39 Vinueza Veloz MF, Zhou K, Bosman LW. et al. Cerebellar control of gait and interlimb coordination. Brain Struct Funct 2015; 220: 3513-3536 doi:10.1007/s00429-014-0870-1
• 40 Keklicek H, Cetin B, Salci Y. et al. Investigating the dynamic plantar pressure distribution and loading pattern in subjects with multiple sclerosis. Mult Scler Relat Disord 2018; 20: 186-191 doi:10.1016/j.msard.2018.01.023
• 41 Kalron A, Pasitselsky D, Greenberg-Abrahami M. et al. Do textured insoles affect postural control and spatiotemporal parameters of gait and plantar sensation in people with Multiple Sclerosis?. PMR 2015; 7: 17-25 doi:10.1016/j.pmrj.2014.08.942
• 42 Korn T. Pathophysiology of multiple sclerosis. J Neurol 2008; 255-6: 2-6 doi:10.1007/s00415-008-6001-2
• 43 Erdeo F, Salcı Y, Uca AU. et al. Examination of the effects of coordination and balance problems on gait in ataxic multiple sclerosis patients. Neurosciences (Riyadh) 2019; 24: 269-277 doi:10.17712/nsj.2019.4.20190038
• 44 Lord SR, McLean D, Stathers G. Physiological factors associated with injurious falls in older people living in the community. Gerontology 1992; 38: 338-346 doi:10.1159/000213351
• 45 Comber L, Galvin R, Coote S. Gait deficits in people with multiple sclerosis: a systematic review and meta-analysis. Gait Posture 2017; 51: 25-35 doi:10.1016/j.gaitpost.2016.09.026
• 46 Altun H, Polat G. On the comparison of classifiers’ performance in emotion classification: Critiques and suggestions. in 2008 IEEE 16th Signal Processing, Communication and Applications Conference. 2008. IEEE