Search

BACKGROUND: The gait modification strategies Trunk Lean and Medial Thrust have been shown to reduce the external knee adduction moment (EKAM) in patients with knee osteoarthritis which could contribute to reduced progression of the disease. Which strategy is most optimal differs between individuals, but the underlying mechanism that causes this remains unknown. RESEARCH QUESTION: Which gait parameters determine the optimal gait modification strategy for individual patients with knee osteoarthritis? METHODS: Forty-seven participants with symptomatic medial knee osteoarthritis underwent 3-dimensional motion analysis during comfortable gait and with two gait modification strategies: Medial Thrust and Trunk Lean. Kinematic and kinetic variables were calculated. Participants were then categorized into one of the two subgroups, based on the modification strategy that reduced the EKAM the most for them. Multiple logistic regression analysis with backward elimination was used to investigate the predictive nature of dynamic parameters obtained during comfortable walking on the optimal modification gait strategy. RESULTS: For 68.1 % of the participants, Trunk Lean was the optimal strategy in reducing the EKAM. Baseline characteristics, kinematics and kinetics did not differ significantly between subgroups during comfortable walking. Changes to frontal trunk and tibia angles correlated significantly with EKAM reduction during the Trunk Lean and Medial Thrust strategies, respectively. Regression analysis showed that MT is likely optimal when the frontal tibia angle range of motion and peak knee flexion angle in early stance during comfortable walking are high (R2Nagelkerke = 0.12). SIGNIFICANCE: Our regression model based solely on kinematic parameters from comfortable walking contained characteristics of the frontal tibia angle and knee flexion angle. As the model explains only 12.3 % of variance, clinical application does not seem feasible. Direct assessment of kinetics seems to be the most optimal strategy for selecting the most optimal gait modification strategy for individual patients with knee osteoarthritis. ispartof: GAIT & POSTURE vol:102 pages:1-9 ispartof: location:England status: published

Achtergrond - De progressie van knieartrose lijkt gerelateerd te zijn aan een hoog extern knieadductiemoment (EKAM), dat gereduceerd kan worden door een aangepast gangpatroon. Doel - Ten eerste is onderzocht welke van vier loopstrategieën het EKAM maximaal verlaagt ten opzichte van comfortabel gaan. Vervolgens is vastgesteld of een maximale EKAM-verlaging altijd wordt gerealiseerd door dezelfde loopstrategie, en of dit kan worden verklaard op basis van de mate waarin de houdingsaanpassing plaatsvindt. Methode - Bij 37 gezonde proefpersonen werd een 3D-ganganalyse gedaan. Na het registreren van het normale gangpatroon werden vier loopstrategieën geïnstrueerd en geregistreerd (Trunk Lean [TL], Medial Thrust [MT], Reduced Vertical Acceleration [RVA] en Toe Out [TO]). Het EKAM en diverse strategiespecifieke kinematica werden berekend voor alle condities. Resultaten - Alle loopstrategieën reduceerden de algehele piek van het EKAM. Hierbij bleken TL en MT de EKAM-piek het meest te reduceren, respectievelijk -35% en -30%. Ook de EKAM-impuls was significant verlaagd door deze strategieën. TO reduceerde de EKAM-puls in de late standfase significant. Er waren duidelijke individuele verschillen zichtbaar. MT verlaagde het EKAM het meest in 43% van de deelnemers, terwijl bij 49% van de deelnemers TL tot de meeste reductie leidde. Hoewel de instructies in beide groepen leidden tot vergelijkbare aanpassing van de kinematica, was de reductie van het EKAM bij uitvoering van dezelfde strategie significant verschillend tussen deze twee groepen. Interpretatie - Hoewel TL en MT het meest effectief waren om het EKAM te reduceren, lijkt individuele selectie van de meest optimale loopstrategie essentieel.

Symptomatic knee osteoarthritis is one of the most common types of osteoarthritis and is one of the top ten causes of years of life lost due to disability. One common treatment strategy involves the prolonged use of anti-inflammatory medications. Recently, gait retraining has been proposed as non-invasive and non-pharmacological treatment option for knee osteoarthritis. However, many possible gait retraining strategies exist and it is unknown a priori which strategy will work best for a given patient. Hence, it is often necessary to pursue a trial-and-error approach to find the best strategy. In this paper we investigate using two standard machine learning techniques, decision trees and rule sets, to build models based on features of a subject’s normal gait to predict which strategy will work best. We were able to learn reasonably accurate models for this task. Furthermore, as the learned models are interpretable, a domain expert was able to gain several insights into the problem by inspecting them. This work shows that machine learning can be useful for prediction of treatment strategies in rheumatology. ispartof: Workshop on Machine Learning in Life Sciences, ECML PKDD 2015 location:Porto date:7-11 September 2015 status: published