| 摘要: | 背景：Fugl-Meyer動作量表(FM)是目前最廣泛使用且完整的中風動作評估工具之一，但其題目多達50題且施測時間長，容易造成評估者與受測者負擔，Fugl-Meyer電腦適性測驗(CAT-FM )與Fugl-Meyer機器學習測驗(FM-ML)提高評估效率，改善評估者與受測者負擔的問題，但兩者的心理計量未得到充分驗證。目的:以FM的UE/LE part(及為FM-UE/LE)為黃金標準，驗證與比較CAT-FM與FM-ML之心理計量、效率。方法：於台北某區域教學醫院招募20歲以上亞急性期與慢性期中風病人個別進行2次評估，使用的評估工具包含Fugl-Meyer動作量表、CAT-FM與FM-ML，驗證與比較CAT-FM、FM-ML之同時效度、再測信度、反應性與效率。結果: CAT-FM與FM-ML在前測與後測都有良好同時效度(r=0.83-0.92，p < 0.001；r=0.90-0.97，p < 0.001)。FM-UE/LE、CAT-FM與FM-ML具有良好再測信度(ICC=0.97-0.99；ICC= 0.92-0.97；ICC= 0.98-0.99)。在團體層級反應性，FM-UE/LE具有中度到高度反應性(SRM=0.63-0.90)，CAT-FM具有中度反應性(SRM=0.53-0.59)，FM-ML具有低度到中度反應性(SRM=0.46-0.70)。在個別層級反應性，總面向為顯著改善組中FM-UE/LE較CAT-FM、FM-ML好(顯著改善人數比:FM-UE/LE =41.93%；CAT-FM =12.90%；FM-ML=12.90%)。在效率方面，FM-UE/LE平均評估時918.12秒，題數為50題；CAT-FM平均評估時間144.91秒，評估題數約4題；FM-ML平均評估時間413.11秒，題數為10題。結論:本研究發現CAT-FM與FM-ML具有良好同時效度、再測信度、中度團體層級反應性與的評估工具，且所需的評估時間、題數較少，可提供研究人員或是臨床醫事人員快速、精確且便利的中風病人動作評估工具的選擇。本研究比較的心理計量共有五項(同時效度、再測信度、團體層級反應性與個別層級反應性、效率)，其中CAT-FM心理計量較FM-ML好的有兩項，分別為團體層級反應性與效率；FM-ML心理計量較CAT-FM好的有三項，分別為同時效度、再測信度與個別層級反應性。CAT-FM適合偵測團體層級病人動作改變與快速評估的選擇，FM-ML適合重複評估與偵測個別層級病人動作改變。預期結果與貢獻：本研究希望透過比較CAT-FM與FM-ML之心理計量、效率，提供研究人員、臨床人員及潛在使用者更多選擇的資訊與參考表準，亦可提供高效率、精準且具備良好心理計量的評估工具以利推廣到中風病人動作評估的普及與正確。
Background: The Fugl-Meyer Assessment(FM-UE/LE) is the most commonly used motor function scale by researchers. Due to the huge amount of items and long conducting process, it is easy to cause the burden on the assessors and the participants. The Fugl-Meyer Computerized Adaptive Test (CAT-FM) and Fugl-Meyer Machine Learning Test(FM-ML) can improve the efficiency of assessment and ameliorate the problem of burdening assessors and participants, however, the psychometrics of both have not been fully validated.
Purposes: This study comparison of the Psychometric Properties of the Fugl-Meyer Computerized Adaptive Test and Fugl-Meyer Machine Learning Test using the gold standard Fugl-Meyer Assessment.
Methods: Recruit subacute stage patients and chronic stage patients over 20 years old in a regional teaching hospital in Taipei for 2 evaluations. The evaluation tools used include the Fugl-Meyer Assessment, the Fugl-Meyer Computerized Adaptive Test, and the Fugl-Meyer Machine Learning Test. Comparison of the concurrent validity, test-retest reliability, responsiveness, and efficiency of the Fugl-Meyer Computerized Adaptive Test and the Fugl-Meyer Machine Learning Test.
Results: The results of CAT-FM and FM-ML both showed good concurrent validity in pre-test and post-test (r=0.83-0.94, P < 0.001; r=0.90-0.97, P < 0.001). Furthermore, FM-UE/LE, CAT-FM and FM-ML all showed high confident results in the test-retest reliability (ICC=0.97-0.99; ICC=0.92-0.97; ICC=0.98-0.99). At group-level responsiveness, the result of FM-UE/LE indicated that it had moderate to high effect size (SRM 0.63-0.90). Moreover, the result of CAT-FM had moderate effect size (SRM 0.53-0.59) and FM-ML had low to moderate effect size (SRM0.46-0.70). Based on the outcomes, the total score was significantly improved in the group FM-UE/LE was better than CAT-FM and FM-ML (significantly improved population ratio: FM-UE/LE = 41.93%; CAT-FM = 12.90%; FM -ML=12.90%). In terms of efficiency, the average evaluation time of FM-UE/LE was 918.12 seconds, and the number of items was 50; the average evaluation time of CAT-FM was 144.91 seconds, and the number of evaluation items was 4; the average evaluation time of FM-ML was 413.11seconds, and the number of items was 10.
Conclusions: This study found that CAT-FM and FM-ML had good concurrent validity, test-retest reliability, moderate group-level responsiveness and evaluation tools, and required less evaluation time and number of items, which may provide researchers and offer a choice with quick, accurate and convenient movement assessment tool for stroke patients for clinical medical personnel; CAT-FM is suitable for detection of group patient motor changes and quick assessment options, FM-ML is suitable for repeated assessment and detection of individual patient motor changes.
Clinical Implementation: This study hopes to comparison of the Psychometric Properties of the Fugl-Meyer Computerized Adaptive Test and Fugl-Meyer Machine Learning Test to prove that they are efficient and accurate assessment tools, so as to facilitate the popularization and accuracy of motor assessment in stroke patients. |
