Abstract
This paper compares the frontal plane hip function of subject’s known to have had hip arthroplasty via either the lateral (LA) or posterior (PA) surgical approaches and a group of subjects associated with no pathology (NP). This is investigated through the Trendelenburg test using 3D motion analysis and classification. Here, a recent development on the Classification and Ranking Belief Simplex (CaRBS) technique, able to undertake n-state classification, so termed NCaRBS is employed. The relationship between post-operative hip function measured during a Trendelenburg Test using three patient characteristics (pelvic obliquity, frontal plane hip moment and frontal plane hip power) of LA, PA and NP subjects are modelled together. Using these characteristics, the classification accuracy was 93.75% for NP, 57.14% for LA, 38.46% for PA. There was a clear distinction between NP and post-surgical function. 3/6 LA subjects and 6/8 PA subjects were misclassified as having NP function, implying that greater function is restored following the PA to surgery. NCaRBS achieved a higher accuracy (65.116%) than through a linear discriminant analysis (48.837%). A Neural Network with two-nodes achieved the same accuracy (65.116%) and as expected was further improved with three-nodes (69.767%). A valuable benefit to the employment of the NCaRBS technique is the graphical exposition of the contribution of patient characteristics to the classification analysis.
Highlights
The considered problem throughout this paper is total hip arthroplasty (THA), and is concerned withAddress correspondence to Gemma Marie Whatling, Cardiff School of Engineering, Cardiff University, Queen’s Buildings, The Parade, Cardiff CF24 3AA, UK
This paper describes the THA problem, the n-state classification and ranking belief simplex (NCaRBS) technique and its application in relation to the THA problem, including emphasis on the elucidation of the contribution of Trendelenburg stance characteristics in the describing evidence they offer in classifying subjects to the three surgery types
The three simplex plots presented show subjects’ predicted associations in the form of their respective simplex coordinates, grouped by their known actual surgery type (LA, posterior approach (PA) or no pathology (NP)), with the grey shaded regions showing, for an object, where there would be predominance to the correct surgery type
Summary
The considered problem throughout this paper is total hip arthroplasty (THA), and is concerned withAddress correspondence to Gemma Marie Whatling, Cardiff School of Engineering, Cardiff University, Queen’s Buildings, The Parade, Cardiff CF24 3AA, UK. The NCaRBS, is a non-parametric evidence-based technique, able to classify objects to a number of states (rather than just two states when using CaRBS4,5,7), using the evidence from a series of characteristics describing them. With DST, the basis of its operations is formulated around the formation of bodies of evidence (BOEs), made up of mass values representing the levels of exact belief (mass values) in associated focal elements Their (BOE) construction, subsequent combination and representation are at the heart of the operations in the NCaRBS, technique. The characteristic-based evidence in the classification process is expressed through the construction of constituent BOEs from an object’s characteristic values vi,j (ith object, jth characteristic), to discern between the object’s association to a state (say {dh}), its complement ðnot dh À f:dhg fd[1]; d2; . The construction of a constituent BOE, defined mi,j,h(Æ) (ith object, jth characteristic, hth state), discerning between {dh}, {Ødh} and {dh, Ødh}, can be best described by reference to Fig. 1
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
