Differences in Clinical Reasoning Structures Based on Experience: A Comparative Network Analysis of Physiotherapists’ Assessment Statements

Background: In Japan, clinical reasoning is often evaluated through case reports. However, the structuring of clinical reasoning has not advanced, making it difficult to objectively compare the validity of descriptions or differences among physiotherapists. This study defines case reports and similar texts as “assessment statements” and extracts and visualizes the structure of thinking processes from them. Methods: We created a network representing the logical structure of assessment statements based on insights from informatics, with “measurement and evaluation” and “phenomenon” as nodes and their causal relationships as edges. Using these networks, we created a system that enables comparing multiple physiotherapists’ thought processes. Two groups, less-experienced (1–3 years, n=9) and more-experienced (5–13 years, n=8), were presented with simulated patient data after femoral transverse fracture surgery and collected assessment statements. Using networks created from these assessment statements, we calculated the average number of nodes (N̄), the average number of edges (Ē), and the edge commonality rate (Ć). These metrics were analyzed to evaluate reasoning-process commonalities based on experience. Additionally, three certified physiotherapists qualitatively verified the network’s utility. Results: Average ± SD and commonality rate values were as follows: a less-experienced group, N̄ = 9.3±3.5, Ē = 7.2±2.9, Ć = 10.9%; a more-experienced group, N̄ = 7.4±3.7, Ē = 6.6±5.2, Ć = 17.5%. Evaluations conducted by certified physiotherapists using the system suggested that the visualized branching structures and unmentioned items enabled the identification of individuals' biases and differences in focus points. Conclusions: A less-experienced group had higher N̄ and Ē values but lower Ć values than the more-experienced group. Less-experienced physiotherapists may merely list a variety of examination items and multiple findings derived from them, suggesting they may not be sufficiently selective in choosing the necessary information. Furthermore, this system demonstrated the ability to objectively compare differences in thought processes that were previously difficult to grasp. Clinical Implications: This study examined a method for representing the structure of clinical reasoning on a network. This enables visualization of individual physiotherapists’ cognitive biases and differences in focus points, thereby supporting shared understanding among physiotherapists and supervisors.