以頻繁樣式樹資料探勘技術評估使用者身心健康指標之研究

由於生活壓力的增加和步調的緊湊，憂鬱症人口不斷向上攀升，憂鬱症的診斷與治療已是不容忽視的一環。憂鬱症的嚴重程度是作為精神科門診前的一項重要參考因素，傳統的門診服務前是採用紙本問卷作答方式讓患者填寫，再由專業助理進行後續的統計分析，因此無法立即判斷憂鬱症程度。本論文設計一身心健康評估診斷系統，包括憂鬱症、焦慮症、睡眠品質、兒童活動量分析和酒癮篩檢。患者在門診前可直接在系統上作答，填寫完後可立即獲得評估結果。患者作答內容可上傳至雲端資料庫儲存，本論文採用頻繁樣式樹以及頻繁樣式增長演算法探勘出患者作答結果間的所有常見項目集，依照關聯法則找出符合的常見項目集。患者的作答結果可立即與資料探勘結果進行比對，提供精神科醫生作出更準確的診斷。針對25,534筆憂鬱症患者的實際作答資料，關聯法則之分數級距在3~5、最小支持度在10~30%及最低信心水準在50~80%時，模擬結果顯示，本研究所採用的資料探勘技術可以探勘出常見項目間以及患者作答間之關聯法則。此外，針對分別由亂數產生作答資料的焦慮症、睡眠品質以及酒癮篩檢，模擬之探勘結果亦顯示可提供醫師判斷精神疾病程度之參考。

關鍵字

憂鬱症；身心健康評估；資料探勘；關聯法則

並列摘要

Due to the pressure from daily life, there is an increase in depression population. The diagnosis and treatment of depression is indispensible for patients. The severity of depression is an important reference factor before psychiatric outpatient service. But the traditional questionnaire is answered by pen on the papers. Then, the responded questionnaire is analyzed by professional assistants. In this way, we could not determine the depression degree immediately. This study is aimed on designing a physical fitness and mental health evaluation system that includes depression, anxiety disorder, sleep quality, children's physical activity and alcohol screening. Patients can answer questionnaire and browse the results directly on the system before outpatient service. The responded questionnaire and analyzed results are automatically uploaded and saved to the cloud database. This study applies frequent pattern trees (FP-tree) and frequent pattern growth (FP-growth) algorithms to discover all the common associations among the records. A new patient's responded questionnaire can be compared with the data mining results that can provide psychiatrists with valuable information to make more accurate diagnosis. This research analyzed the association rules from combinations of various grade intervals of 3~5, minimum supports of 10~30% and minimum confidence of 50~80% on the actual 25,534 records in our database. The result shows that the proposed system can find frequent itemsets and interesting association rules from databases. In addition, the simulation results from randomly generated data of anxiety disorder, sleep quality, and alcohol screening are also valuable references for psychiatrists to diagnose psychiatry.

並列關鍵字

Depression ； Physical and Mental Health Evaluation ； Data Mining ； Association Rules

參考文獻

[13] 鄧政宏，基於視覺式多點觸控系統之使用者觸控模式分析，碩士論文，國立臺北科技大學電機工程研究所，臺北，2011。

[2] 行政院衛生署國民健康局，http://www.bhp.doh.gov.tw/BHPnet/Portal/

[34] 陳彥良、趙書榮、陳禹辰，“幾個快速挖掘關聯規則的資料探勘方法”，電子商務學報第5卷第2期，民國92年6月。

[32] X. Shang, K. Sattler and I. Geist, “SQL based frequent pattern mining without candidate generation,” in Proc. of the 2004 ACM Symp. Conf. on Applied Computing, Nicosia, Cyprus, pp.618-619, March 2004.

[5] D. Perera, J. Kay, I. W. Koprinska and K. Yacef, “Clustering and sequential pattern mining of online collaborative learning data,” IEEE Trans. on Knowledge and Data Engineering, vol. 21, no. 6, pp.759-772, June 2009.

國際替代計量

以頻繁樣式樹資料探勘技術評估使用者身心健康指標之研究

未授權

主題瀏覽