Clustering of shoulder movement patterns using K-means algorithm based on the shoulder range of motion

Objectives: This study aimed to classify and identify shoulder movement patterns based on shoulder joint range of motion (RoM) by applying the K-means clustering algorithm.

Design: Observational study using data from the 5th Size Korea Anthropometric Survey (2003-2004).

Setting: Data analysis focused on anonymized shoulder RoM measurements from a national survey.

Participants: Analysis included 541 participants after excluding those with incomplete shoulder RoM data.

Main outcome measures: Identification of clusters based on measurements of shoulder flexion, extension, internal rotation, external rotation, horizontal adduction, and horizontal abduction.

Results: Eight distinct clusters were identified, each showing unique shoulder mobility characteristics. Clusters 1 and 5 had the lowest flexion ranges, whereas clusters 7 and 8 exhibited low internal rotation and horizontal adduction. Clusters 2 and 6 displayed the highest flexion and overall high flexibility, while clusters 3 and 4 presented moderate flexion with low horizontal adduction.

Conclusions: This observational study categorized shoulder movement into eight clusters, revealing diverse mobility patterns across the general population. This clustering provides a basis for potential research into the correlation between specific movement patterns and musculoskeletal disorders, aiding in the development of targeted therapeutic strategies.