職能治療師對身心障礙兒童依職能治療所定義42評估選項進行類群評估，釐清個案問題歸屬以提出更有效復健治療建議，適切安排短期及長期治療計劃，期能恢復個案自主生活的能力。本研究針對彰化基督教醫療體系員生醫院復健科224份個案評估報告，利用資料探勘中K均值法對資料進行類群的分屬，進而獲得各類群重要特徵因子；再採用決策樹演算法針對特徵因子是否能具有高預測效能進行評估。研究結果發現，K均值法將個案分三類群，分別是生活自理型、肢體動作型及感覺處理調節型，這結果與Smith&Brien於1996職能復健兒童所區分的三大領域相符，並篩選出20項重要因子。各因子異常屬性則可為各類群參考特徵項，如社會互動、前庭本體覺處理、和姿勢動作相關的感覺調節、粗大與精細動作、理解及表達能力、進食、盥洗衛生及穿脫衣物。運用CART決策樹演算法驗證因子預測準確率可達100%，規則集中各規則預測準確率也達100%的信任度。由此可見，資料探勘技術確實能由職能治療領域複雜資料中，整理出重要且具有特徵又有代表性因子，讓職能治療師能有效為身心障礙兒童對症下藥。

The occupational therapist needs to evaluate 42 items of the evaluation report in order to understand the indeed problems of the children in occupational therapy. Meanwhile, the proper rehabilitation arrangement would be prepared for the children after identifying their problems and affects that children perform their independence and responsibility on self-caring, learning, and playing. This study used the K-Means techniques to classify the problems of the children from the 224 evaluation reports and the important features which belonged would be extracted. Then the decision tree algorithm was used to predict the performance of the clustering model. The results of this study showed that 3 clusters was classified which were problems of daily life, body motor, and sensory processing. Twenty important features were extracted and presented for the problems. The unusual attributes of each problem will be reference items for the therapist, such as, socialization, proprioceptive and vestibular, postural control related, gross and fine coordination, comprehension and expression, feeding and eating, bathing and showering, and dressing. The model of decisions trees were established by the 20 important features and the accuracy was 100%. The rule set achieved 100% confidence in particular samples. The results with 3 clusters of this study confirm the definition of occupational rehabilitation.