Predictors affecting balance performances in patients with knee osteoarthritis using decision tree analysis

Abstract

Purpose: Knee osteoarthritis (OA) is one of the most predominant causes of pain, functional decline and disability in the elderly population worldwide. Individuals with knee OA typically exhibit knee pain, impaired range of motion, and weakness (particularly in the quadriceps muscles) compared with those without knee OA. These impairments may be related to reduce physical activity, and may combine to explain the poorer balance observed in this patient population. Swinkels et al reported that 24 % of patients with severe knee OA experienced a fall prior to surgery. It is important to identify patient and functional characteristics that contribute to poor balance. However, we are unaware of any studies that have examined functional predictors of balance performance in those with knee OA. Also, no quantifiable measures are available to indicate that a patient is at risk for a poorer functional outcome. Therefore, we developed a preliminary decision algorithm predicting functional performance using joint performance measures, anthropomorphic measures, and demographic measures. This algorithm utilizes simple, quantifiable clinical measures that could be used to help to consider the physical therapy for patients with knee OA. A purpose of this study is to investigate related factors to balance ability in patients with knee OA using decision tree analysis. Methods: Thirty three patients with knee OA (32 women, a man, mean age 68.5 ± 7.3 years; height 150.8 ± 5.2cm; weight 60.6 ± 8.0kg) participated in this study. Each subject gave written informed consent before entering this study. Anthropometry, severity in gonarthritis, femorotibial angle (FTA), knee pain level, knee range of motion (ROM), knee muscle strength, one leg standing (OLS) test, functional reach (FR) test, timed up and go (TUG) test were assessed. Kellgren and Lawrence (KL) grade was used to determine severity of gonarthritis. FTA was defined as the lateral angle formed by the femur and tibia. Lines were drawn through the middle of the femoral shaft and through the middle of the tibial shaft. Knee pain level during OLS, standing-up, and walking were evaluated by a visual analog scale (VAS). Knee flexion and extension ROM were measured passively in bilateral limbs using goniometer. Knee isokinetic strength was assessed bilaterally using the Biodex system 3 (Biodex Medical System Inc.). The isokinetic protocol measures knee flexion and extension peak torque of concentric contraction at an angular velocity of 180 degrees / sec. For evaluation of balance performance, OLS time, FR distance, and TUG time were measured. Sex, age, height, weight, BMI, KL grade, FTA, knee pain level, knee ROM, and knee strength were evaluated as predictors of balance performance. Decision tree analysis was used to determine the variables that related to balance performance. We developed a decision algorithm using Classification and Regression Trees (CART). The level of significance was set at 0.05. Results: Mean ± SD performance on the OLS time for patients was 22.8 ± 19.6 s. The p...

Identification of Risk Factors and Machine Learning-Based Prediction Models for Knee Osteoarthritis Patients

“…The interpretation of the weights in logistic regression differs from the interpretation of the weights in linear regression, since the outcome in logistic regression is a probability between 0 and 1. We also evaluated decision trees (DTs) [27] which are a non-parametric supervised learning method used for classification and regression. They are simple to understand and to interpret.…”

Applications of machine learning algorithms in quality of life

“…The interpretation of the weights in logistic regression differs from the interpretation of the weights in linear regression, since the outcome in logistic regression is a probability between 0 and 1. We also evaluated decision trees (DTs) [155] which are a non-parametric supervised learning method used for classification and regression. They are simple to understand and to interpret.…”