Novel approach for spastic hemiplegia classification through the use of support vector machines

Abstract

The use of computer based classification methods for spastic hemiplegia (SH) is a relatively new approach. Before and still, the classification is primarily done by specialized physicians through manual template matching on kinematics plots of the sagittal plane data from the affected limb generated by gait laboratories. In the past, several methods for computer based automation of this classification have been attempted. This paper introduces the use of support vector machine (SVM) as a model that contributes to this process. The results obtained from the use of SVM are quite efficient considering the data set utilized involves patients from an array of ages and both sex. It should be mentioned that even though the number of patients is not trivial; the percentages of accurate classification represented are promising. Since the SVM method improves its accuracy with an increased number of training cases, this approach has the advantage of becoming more accurate with time. Many more significant discoveries are expect from the introduction of SVM in the analysis of SH and further studies of gait related pathologies are expected to contribute as well.

Gait classification in children with cerebral palsy: A systematic review

“…It is likely that these studies used a sample of convenience because this type of sampling involves the selection of the most accessible members of the target population [40]. Five studies [2,7,10,18,19] reported using a retrospective sample of convenience chosen from children who had already attended a health service such as a gait laboratory or orthopaedic outpatient unit. The main limitation of this sampling method is that the sample may be biased and not representative of the entire population of interest [40].…”

“…Cross-validation attempts for the classifications were conducted by four studies [2,8,10,16]. The purpose of this cross-validation process was to assess the precision of the classification system by using a test set of data that was not from the original set of data used to construct the classification.…”

“…These included a generalised dynamic neural network [19], Hidden Markov Models [2] and the 'Support Vector Machine' [10]. These types of techniques or models use computing algorithms that attempt to imitate and automate complex pattern recognition tasks much like the human brain [51].…”

“…The diversity of gait deviations observed in children with CP has led to repeated efforts to develop gait classification systems to assist in diagnosis, clinical decision-making and communication [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. Gait classifications may enable clinicians to differentiate gait patterns into clinically significant categories that assist with clinical decision-making.…”

The validity of an existing gait classification system when applied to a representative population of children with hemiplegia

“…Of the gait classification systems proposed for spastic hemiplegic CP [2][3][4][5], the Winters' Classification [2], based upon sagittal plane gait patterns and Perry's premise that recovery proceeds from proximal to distal end of the limb, seems to be the most widely accepted within the gait analysis community. It is of particular clinical relevance as it describes the level of involvement for a particular patient.…”