Development of an effective clustering algorithm for older fallers

Abstract

Falls are common and often lead to serious physical and psychological consequences for older persons. The occurrence of falls are usually attributed to the interaction between multiple risk factors. The clinical evaluation of falls risks is time-consuming as a result, hence limiting its availability. The purpose of this study was, therefore, to develop a clustering-based algorithm to determine falls risk. Data from the Malaysian Elders Longitudinal Research (MELoR), comprising 1411 subjects aged ≥55 years, were utilized. The proposed algorithm was developed through the stages of: data pre-processing, feature identification and extraction with either t-Distributed Stochastic Neighbour Embedding (t-SNE) or principal component analysis (PCA)), clustering (K-means clustering, Hierarchical clustering, and Fuzzy C-means clustering) and characteristics interpretation with statistical analysis. A total of 1279 subjects and 9 variables were selected for clustering after the data pre-possessing stage. Using feature extraction with the t-SNE and the K-means clustering algorithm, subjects were clustered into low, intermediate A, intermediate B and high fall risk groups which corresponded with fall occurrence of 13%, 19%, 21% and 31% respectively. Slower gait, poorer balance, weaker muscle strength, presence of cardiovascular disorder, poorer cognitive performance, and advancing age were the key variables identified. The proposed fall risk clustering algorithm grouped the subjects according to features. Such a tool could serve as a case identification or clinical decision support tool for clinical practice to enhance access to falls prevention efforts.

A Clustering-Based Approach to Functional and Biomechanical Parameters Recorded with a Pair of Smart Eyeglasses in Older Adults in Order to Determine Physical Performance Groups

“…An unsupervised analysis was performed to determine the optimal number of clusters and to allocate the participants to the different groups of physical performance. First, a clustering algorithm was created in Python (version 2.7.18) based on K-means (unsupervised analysis) [ 27 ]. Second, the optimal number of clusters was identified from the maximum silhouette coefficient [ 39 ].…”

“…For cluster determination, the initial centroids in each group were selected randomly, and each participant was assigned to the closest centroid by calculating the Euclidean distance. The initial positions were refined during the analysis to create centroids that were the most spaced in terms of distance from each other, and to have the smallest distance between participants in the same group to centroids [ 27 ]. Participants were then assigned to different distinct groups of physical performance based on the coefficient score and a performance equation was obtained through linear regression of the functional and biomechanical parameters.…”

“…Due to the complexity and non-linearity that characterizes the clinical status of individuals, an alternative approach was recently used to categorize the frailty or faller status in older adults from multifactorial measures [ 26 , 27 ]. This approach is based on unsupervised clustering analysis that aims at identifying similarities in clinical outcomes within a population.…”

“…For example, blood analyses (e.g., total cholesterol, fasting glucose, hemoglobin, hematocrit) and both waist and arm circumferences were measured based on their capacity to better categorize frailty in older adults [ 26 ]. Similarly, the categorization of the risk of falling was determined from different clinical tests (i.e., TUG, functional reach test, and handgrip strength) and cardiovascular data (e.g., systolic and diastolic blood pressure variability ratio and heart rate variability ratio) [ 27 ]. The authors obtained four groups corresponding to physical performance levels that were associated with different levels of fall risk.…”

“…The authors obtained four groups corresponding to physical performance levels that were associated with different levels of fall risk. Considering the non-linearity observed in frailty development [ 26 ], as well as faller status [ 27 ], this approach seems more relevant to categorizing older adults according to different physical performance levels.…”