Rozumalski, Adam2016-08-192016-08-192016-04https://hdl.handle.net/11299/181657University of Minnesota Ph.D. dissertation. April 2016. Major: Biomedical Engineering. Advisor: Michael Schwartz. 1 computer file (PDF); vi, 102 pages.Objectives Clinical gait analysis, as commonly prescribed for children with Cerebral Palsy, is a complex set of procedures which include examining data from several sources. The tools developed with this project will use that data to provide robust, repeatable, evidence-based guidance to highlight the most effective treatments for children with CP. These tools will also supply objective measures that can be included in outcome analysis. Methods Several mathematical techniques are used to find patterns within the gait date including: singular value decomposition of kinematic and kinetic data to measure gait pathology; k-means cluster analysis of those results to find recurring patterns; principal components analysis of physical exam findings to relate the gait patterns to physical function; and non-negative matrix factorization of electromyography data to measure motor control. Results The decomposition and scaling of the kinematic and kinetic data resulted in a set of indexes that are able to quantify gait pathology. The k-means cluster analysis reveals that there are repeatable patterns within the gait pathology. These patterns are related to clinical findings as calculated from principal components analysis. Clinical interpretations of motor control can be quantified as muscle synergies using non-negative matrix factorization. Interpretation These tools have proven to provide important quantitative information on treatment outcomes. When implemented in routine clinical gait analysis, these tools have the ability to provide evidence based guidance in treatment decisions.enCerebral PalsyDecision Support ToolsGait AnalysisNon-negative Matrix FactorizationPrincipal Component AnalysisSingular Value DecompositionThesis or Dissertation