Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/de-gruyter/computational-gait-analysis-using-fuzzy-logic-for-everyday-clinical-7z1s8vNuGR
- Publisher
- de Gruyter
- Copyright
- Copyright © 2017 by the
- ISSN
- 1895-9091
- eISSN
- 1896-530X
- DOI
- 10.1515/bams-2016-0023
- Publisher site
- See Article on Publisher Site
Abstract
Background: Proper, early, and exact identification of gait impairments and their causes is regarded as a prerequisite for specific therapy and a useful control tool to assess efficacy of rehabilitation. There is a need for simple tools allowing for quickly detecting general tendencies. Objective: The aim of this paper is to present the outcomes of traditional and fuzzy-based analysis of the outcomes of post-stroke gait reeducation using the NeuroDevelopmental Treatment-Bobath (NDT-Bobath) method. Materials and methods: The research was conducted among 40 adult people: 20 of them after ischemic stroke constituted the study group, and 20 healthy people constituted the reference group. Study group members were treated through 2 weeks (10 therapeutic sessions) using the NDT-Bobath method. Spatio-temporal gait parameters were assessed before and after therapy and compared using novel fuzzy-based assessment tool. Results: Achieved results of rehabilitation, observed as changes of gait parameters, were statistically relevant and reflected recovery. One-number outcomes from the proposed fuzzy-based estimator proved moderate to high consistency with the results of the traditional gait assessment. Conclusions: Observed statistically significant and favorable changes in the health status of patients, described by gait parameters, were reflected also in outcomes of *Corresponding author: Emilia Mikolajewska, Ludwik Rydygier Collegium Medicum in Bydgoszcz, Nicolaus Copernicus University in Toru, Department of Physiotherapy, Jagielloska 13-15, 86-067 Bydgoszcz, Poland, E-mail: e.mikolajewska@wp.pl; and Centre for Modern Interdisciplinary Technologies, Nicolaus Copernicus University in Toru, Neurocognitive Laboratory, Wileska 5, 87-100 Toru, Poland Piotr Prokopowicz: Institute of Mechanics and Applied Computer Science, Kazimierz Wielki University in Bydgoszcz, Kopernika 1, 85-064 Bydgoszcz, Poland Dariusz Mikolajewski: Institute of Mechanics and Applied Computer Science, Kazimierz Wielki University in Bydgoszcz, Kopernika 1, 85-064 Bydgoszcz, Poland; and Centre for Modern Interdisciplinary Technologies, Nicolaus Copernicus University in Toru, Neurocognitive Laboratory, Wileska 5, 87-100 Toru, Poland fuzzy-based analysis. Proposed fuzzy-based measure increases possibility of the clinical gait assessment toward more objective clinical reasoning based on common use of the mHealth solutions. Keywords: fuzzy-based analysis; gait; neurologic gait disorders; physical therapy modalities; rehabilitation. Introduction Gait is regarded as necessary for mobility and most of the activities of daily living; it is simply essential for overall quality of life. Moreover, walking, as a physical activity, protects against cognitive impairment. Proper walking also reduces the risk of falling. Proper, early, and exact identification of gait impairments and their causes is regarded as a prerequisite for specific therapy and a useful control tool to assess efficacy of rehabilitation. Slow gait with reduced body dynamics is regarded as distinctive feature in the elderly; gait disorders may be also an early predictor of dementia in selected geriatric patients (depending on the type of dementia) [13]. There is a need for simple tools allowing for quickly detecting general tendencies in gait assessment for both early identification of gait disorders and progress of the gait recovery process. Too many too complex tools cause clinical practice to be more expensive but not always quicker, simpler, and more effective. Moreover, there is a need for diagnostic tools easy in remote administration (e.g. in telerehabilitation), semi-automatic analysis, and interpretation. As main applications are regarded the rehabilitation and diagnosis of various medical conditions (medical diagnostics, rehabilitation) and sport activities. Four main clinical applications were identified so far: assessment of movement disorders, assessment of surgical outcomes, improvement of walking stability, reduction of joint loading [4]. There is a common belief that continuous, long-term clinical gait assessment in patient population should take place outside the laboratory, in natural human environment (patient's home, work, community) [4]. Aforementioned monitoring during activities of daily living in the 38Mikolajewska: Clinical gait analysis using fuzzy logic natural environment can increase efficacy of intervention. Thus inexpensive, convenient, and efficient solutions are needed. Such approach still constitutes a huge challenge; quality gait analysis requires a large amount of data and specialized technical solutions designed to gather (real-time data acquisition from multiple sensors), store, manage, and extract clinically relevant information [5]. The aim of this paper is to present the outcomes of traditional and fuzzy-based analysis of the outcomes of post-stroke gait reeducation using the NeuroDevelopmental Treatment-Bobath (NDT-Bobath) method. gait parameters once. Measurement was provided by the same experienced specialist as in the study group to avoid inter-rater errors. Patients' overall profile is presented in Table 1. The data were analyzed with Statistica 12 (StatSoft, USA) software. Results analysis was carried out both for normal and for pathological walking patterns. The results of measurements were given as mean, standard deviation (SD), median, maximal value, and minimal value. Shapiro-Wilk test was used to establish normality of the distribution. The results were statistically analyzed using the Student t-test or Wilcoxon's test. Changes between first and second measurements in the study group were calculated as a result of the subtraction. To assess correlations, Spearman's rank correlation coefficients (Spearman's rhos) were calculated. The level of statistical significance was set at p0.05. Two concurrent versions (called method 1 and Method 2) of the Multicriteria Fuzzy Evaluator of Gait (MuFEG) by Prokopowicz were used to assess the gait parameters. The cumulated measure of gait is presented as one-number outcome: percentage, where 100% is an ideal gait, and results for the reference group (healthy people) cannot be lower than 50%. Values lower than 50% are regarded as pathological. Technically different descriptors of a gait are represented by fuzzy systems of Mamdani type. Assessment was additionally checked using Fuzzy Logic Toolbox for MATLAB for reference purposes. Table 1:Patients' overall profile. Parameter Reference group (n=20) 10 (50%) 10 (50%) 59.30 6.71 65.50 51.00 72.00 n.a. Study group (n=20) 9 (45%) 11 (55%) 64.95 9.70 67.50 49.00 82.00 11 (55%) 9 (45%) Materials and methods Design was prospective before-after study (BAS). The research was conducted among 40 adult patients. Twenty of them who had undergone ischemic stroke constituted the study group. Inclusion criteria were the following: age18 years old, time after stroke of up to 3 years, and ability to walk (even despite pathological walking patterns). Exclusion criteria covered age <18 years old, time after stroke >3 years, inability to walk, and significant secondary changes due to another diseases or injuries influencing the gait pattern. These patients were treated using the NDT-Bobath method one of the most popular therapeutic methods in neurorehabilitation, including post-stroke. Gait reeducation in the study group lasted for 2 weeks (10 therapeutic sessions, each of them lasted half an hour preparation of patient not included). Therapy was carried by the same therapists, International Bobath Instructors Training Association (IBITA) recognized (both basic and advanced course) with 15 years of experience in neurorehabilitation. Each patient was assessed during a 10-m walk test (Figure 1) using spatio-temporal gait parameters, i.e. gait velocity, cadence and stride length, and their normalized values. Measurements of these parameters were conducted by the same person twice: on admission and after the last session of the therapy. Another set of 20 healthy people constituted the reference group. Each member of the reference group was assessed during a 10-m walk test using spatio-temporal Direction of motion Speeding up Start timing 10 m Finish timing Slowing down Sex Male Female Age, years Mean SD Median Min Max Side of paresis Left Right Time after cerebrovascular accident (CVA), years Mean SD Median Min Max n.a., Not applicable. n.a. Figure 1:Ten-meter walk test. Mikolajewska: Clinical gait analysis using fuzzy logic39 The study was accepted by the appropriate Bioethical Committee. The subjects gave written informed consent before entering the study, in accordance with the recommendations of the Bioethical Committee, acting on the rules of Good Clinical Practice and the Helsinki Declaration. Table 2:Spatio-temporal gait parameters. Parameter Reference Study group (n=20) group Before After Change (n=20) therapy therapy 1.81 0.17 1.80 1.60 2.20 123.45 11.47 124 102 142 1.76 0.18 1.67 1.54 2.00 0.60 0.06 0.58 0.52 0.72 0.62 0.06 0.61 0.54 0.73 1.93 0.21 1.84 1.7 2.23 0.46 0.52 0.06 0.18 0.19 0.21 0.50 0.40 0.10 0.10 0.10 -0.20 0.80 1.60 0.80 75.75 78.95 3.25 18.17 24.28 17.50 77.50 76.50 -2.00 36.00 24.00 -36.00 100.00 151.00 67.00 1.41 1.54 0.10 0.38 0.33 0.24 1.48 1.54 0.14 0.57 0.61 -0.55 2.22 2.5 0.5 0.15 0.18 0.03 0.06 0.05 0.03 0.16 0.15 0.01 0.05 0.04 -0.07 0.28 0.53 0.25 0.38 0.39 0.02 0.09 0.17 0.05 0.39 0.38 -0.01 0.17 0.12 -0.18 0.51 0.76 0.34 1.57 1.78 0.16 0.59 0.59 0.36 1.68 1.82 0.17 0.38 0.72 -0.6 2.5 2.81 1.19 12.99 4.11 5.77 0 33.63 19.99 6.94 20.84 0 54.47 6.88 2.11 9.28 0 20.08 Results The results are presented in Tables 24. Discussion Despite recent dramatic development of the sophisticated computational models, current understanding of the human gait is still limited [68]. The biggest challenge constitutes quick, cheap, and valid everyday routine clinical gait analysis. Visual gait analysis is cheap but still inaccurate; the study by Lord et al. showed that complete agreement in scores of Rivermead Visual Gait Assessment (RVGA) occurred only on 63.8% of observations [9]. Thus visual gait analysis should be used rather in conjunction with other, more accurate methods of gait assessment. Clinical reliability and validity of wearable sensors and feedback devices still need further assessment on patients with various diseases impairing gait function, and their impact on human gait is unknown [4]. They are divided into tools assessing gait kinematics, gait kinetics, and electromyography [10]. Current clinical value of gait assessment depends on many factors, including understanding, expectations and preferences of the medical staff, ability to interpret and incorporate results into everyday clinical practice, applicability of the current technical devices to particular clinical problems, previous limited use of traditional tests to address gait disorders in selected medical conditions (due to long time, costs, and limited accessibility to gait laboratories), and application of other clinical tests [8]. Communication between biomedical engineers and clinicians and technical abilities of the medical staff grows slowly [8]. We hope this co-operation will develop and will be more fruitful in the future. Computational modeling and power of artificial intelligence may improve clinical reasoning within the most complex clinical procedures, such as gait reeducation supported by application of rehabilitation robots and exoskeletons [11]. Gait velocity Mean SD Median Min Max Cadence Mean SD Median Min Max Stride length Mean SD Median Min Max Normalized gait velocity Mean SD Median Min Max Normalized cadence Mean SD Median Min Max Normalized stride length Mean SD Median Min Max Fuzzy-based analysis method 1, % Mean SD Median Min Max Fuzzy-based analysis method 2, % Mean SD Median Min Max n.a., Not applicable. 40Mikolajewska: Clinical gait analysis using fuzzy logic Table 3:Correlations for healthy people. Parameter Age Gait velocity Cadence Stride length Normalized gait Normalized Normalized Fuzzy evaluator velocity cadence stride length Method 1 Method 2 -0.952 p=0.000 ns ns -0.397 p=0.082 0.458 p=0.042 -0.620 p=0.004 -0.948 p=0.000 0.956 p=0.000 ns ns -0.452 -0.449 p=0.045 p=0.047 ns 0.518 p=0.019 0.905 -0.516 p=0.000 p=0.020 -0.467 0.922 p=0.038 p=0.000 ns 0.482 p=0.031 ns 0.454 0.003 0.611 p=0.003 0.633 p=0.001 0.357 p=0.004 0.399 p=0.002 0.447 p=0.002 0.523 p=0.000 0.444 0.047 0.722 p=0.002 0.655 p=0.003 0.336 p=0.000 0.394 p=0.041 0.445 p=0.007 0.542 p=0.001 0.936 p=0.000 Age Gait velocity Cadence Stride length Normalized gait velocity Normalized cadence Normalized stride length Method 1 Method 2 Table 4:Correlations among changes for post-stroke patients. Parameter Age Change of Change of Change of Change of Change of Change of Change of fuzzy gait velocity cadence stride length normalized normalized normalized evaluator gait velocity cadence stride length Method 1 Method 2 n.s. n.s. 0.756 p=0.000 n.s. 0.311 p=0.015 n.s. n.s. 0.943 p=0.000 0.745 p=0.000 n.s. n.s. 0.894 p=0.000 0.860 p=0.000 n.s. 0.857 p=0.000 n.s. n.s. n.s. 0.937 p=0.000 n.s. n.s. 0.411 0.030 0.522 p=0.011 n.s. 0.333 p=0.000 0.376 p=0.001 0.434 p=0.008 0.555 p=0.030 0.435 0.005 0.765 p=0.007 0.456 p=0.011 0.378 p=0.006 0.378 p=0.004 0.432 p=0.009 0.532 p=0.022 0.818 p=0.001 Age Change of gait velocity Change of cadence Change of stride length Change of normalized gait velocity Change of normalized cadence Change of normalized stride length Change of method 1 Change of method 2 Multifactorial disorders need multimodal therapies providing simultaneous comprehensive and affective treatment of the specific deficits and aspects of everyday function. Neurological deficits, such as sensory deficits, neurodegeneration, cognitive syndromes, cognitive impairment, degeneration of joints, and sarcopenia, are common causes of gait disorders, causing fear of falling and associated secondary changes due to low physical activity [2]. Achieved results of rehabilitation, observed as changes of gait parameters, were statistically relevant and reflected the recovery (mean change was >0). Despite the low number of evidence concerning results of gait recovery due to NDT-Bobath-based therapy, outcomes of the study confirm and enhance previous studies [12, 13]. Moderate to high negative correlations of age and gait parameters show worsening of the spatio-temporal gait parameters with age in healthy people, as far as lower results of Mikolajewska: Clinical gait analysis using fuzzy logic41 gait neurorehabilitation in older adults. Gait parameters (velocity, cadence, stride length, and their normalized values) are mild to highly positively correlated, i.e. it is hard to achieve recovery of velocity without improvement of cadence and stride length. What is important is that there was an observed moderately negative correlation between cadence and stride length in healthy people. The main limitation of the study is the low number of participants; thus we regard our study as preliminary. Directions for further studies will be two-fold: increase sample size and check method(s) suitability on outcomes from randomized controlled trial (e.g. comparing two rehabilitation methods in randomized sample). Fuzzy logic is relatively rarely used in tools for clinical gait analysis. Studies by Armand et al. [14], Senanayake et al. [5], and Sagawa et al. [15] showed usefulness of fuzzy logic in the aforementioned area. Our study, to our best knowledge, is the first study concerning use of ordered fuzzy numbers (Kosiski's fuzzy numbers) in this application. Presented outcomes and analysis of the computational part of the study show method 2 as more suitable for treated patients and disorders, but it needs additional research. The most important part of the computational tool preparation is not a general algorithm but fine tuning. Switching among fuzzy classifier parameters may be necessary to assess gait in various patients to avoid increased errors or even malfunction. Simplicity and relatively low computational requirements of the proposed method allow to it to be built in a smartphone or tablet and use (with built-in digital camera) as portable gait analysis system, supplementing traditional clinical gait analysis both at the rehabilitation ward (as remote part of Hospital Information System) and at the patient's home (even as a component of telerehabilitation system). Proposed solution is low cost and available for patients with no access to the gait laboratory as far as easy screening tool for general practitioners. We work on the further development of our tool toward application of ordered fuzzy numbers, fractal measurements, and semi-intelligent classification based on neural networks. One-number digit result of the assessment is easy to interpret and use within clinical reasoning. Change of the aforementioned results may be presented in a more informative graphical form needed for further clinical studies. Use of the application is similar to the traditional smartphone apps, thus user-friendly and intuitive. were reflected also in outcomes of fuzzy-based analysis. Proposed fuzzy-based measure increases possibility of the clinical gait assessment toward more objective clinical reasoning based on common use of the mHealth solutions. Further research is needed including both clinical studies (randomized controlled trials on bigger samples) and technical development (work on remote and local part of the system, co-operation within bigger IT environments, automation of the measure, and analysis). Author contributions: The authors have accepted responsibility for the entire content of this submitted manuscript and approved submission. Research funding: None declared. Employment or leadership: None declared. Honorarium: None declared. Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.
Journal
Bio-Algorithms and Med-Systems – de Gruyter
Published: Mar 1, 2017