運用機器學習方法分析急診病患檢傷資訊以預測 X 光檢查必要性

簡易檢索 / 詳目顯示

回結果列表

研究生：	黃信立 Huang, Hsin-Li.
論文名稱：	運用機器學習方法分析急診病患檢傷資訊以預測 X 光檢查必要性 Analyzing emergency patient triage information using machine learning methods to predict X-ray examinations necessity
指導教授：	陳建良 Chen, James C.
口試委員:	陳盈彥 Chen, Yin-Yann 陳子立 Chen, Tzu-Li
學位類別：	碩士 Master
系所名稱：	工學院 - 工業工程與工程管理學系碩士在職專班 Industrial Engineering and Engineering Management
論文出版年：	2022
畢業學年度：	110
語文別：	中文
論文頁數：	61
中文關鍵詞：	急診室、機器學習、X 光檢驗
外文關鍵詞：	machine learning, Emergencydepartment, X-rays
相關次數：	點閱：1 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

急診壅塞是台灣教學級醫院常見的情形，因此改善急診整體流程成為解決方法之一。因病人進入急診第一站是檢傷分類，本研究欲以急診病人於檢傷分類時提供的資料，運用機器學習建立該病人是否需要進行X光檢查的預測模型，可作為醫生診斷時的建議或其他檢驗排程安排的參考。

研究分析自西元 2011 年 1 月到 2020 年 12 月止的台灣某教學級醫學中心的急診病人，擷取其檢傷分類時所提供的資料。我們採取 X 光檢驗與否作為目標，試著依此結合機器學習（邏輯斯迴歸、隨機森林、XGBoost），建立病人是否需要進行 X 光檢查的預測模型。並以其準確率跟接收者操作特徵曲線下的面積 (Area Under the Curve Receiver operating characteristic curve, AUROC)，以推估其表現，並以存活曲線確認此模型是否適用入院 2 小時後病人是否接受X光的預測。

本資料搜集 357,739 位病患，資料整理後，使用 306,736 位病患資料，分為三組做訓練與測試；以檢傷分類專業用語作為新增特徵後，帶入 XGBoost 模型訓練。在以梯度下降法進行超參數優化下，能達到最佳的接收者操作特徵曲線下面積 (AUROC): 0.754 以及準確率 0.704。該模型對於台灣急診新冠肺炎前後的測試集結果差距不大。帶入預測模型得到的存活曲線也呈現顯著差異。

Emergency room overcrowding is common in teaching hospitals in Taiwan. Therefore, improving the overall emergency department procedure is one of the solutions. As the first station in the emergency department is triage, this study aims to provide information for triaging of emergency department patients. Machine learning was used to construct a predictive model to determine if the patient requires X-ray examination, which provides a reference for physician diagnosis or arrangement of other examinations.

This study analyzed emergency department patients from a teaching medical center in Taiwan from January 2011 to December 2022 and extracted triage information. We used X-rays as a target and attempted to employ machine learning (logistic regression, random forest, XGBoost）to construct a predictive model to determine if the patient requires Xray examination. The accuracy and area under the receiver operating characteristic (AUROC) curve were used to evaluate the performance of the model and a survival curve was used to confirm if this model is suitable for prediction of whether patients require Xrays 2 hours after admission.

We collected 357,739 patients for the datasets. After the data was organized, 306,736 patient datasets were used and divided into 3 groups for training and testing. Triagespecific terms were used as a new characteristic and introduced into the XGBoost model for training. The gradient descent method was used for hyperparameter optimization to obtain the optimal AUROC and accuracy, which were 0.754 and 0.704, respectively. The differences in test results of this model before and after the COVID-19 pandemic in Taiwan were not significant. The survival curve obtained by introducing the predictive model also showed significant differences.

摘要 ..................................................................................................................................... I Abstract .............................................................................................................................. II

致謝 ...................................................................................................................................IV 目錄 .................................................................................................................................... V

表目錄 ............................................................................................................................. VII 圖目錄 ............................................................................................................................ VIII

第一章 緒論 ....................................................................................................................... 1

1 背景和研究動機 ................................................................................................. 1

1.1 急診流程與檢傷分類 ..................................................................................... 2

1.2 新冠肺炎與台灣急診 ..................................................................................... 4

2 研究目標 ............................................................................................................. 4

3 論文架構 ............................................................................................................. 5

第二章 文獻探討 ................................................................................................................ 6

1 改善急診壅塞的方法 ......................................................................................... 6

2 人工智慧與急診預測 ......................................................................................... 8

第三章 研究方法 .............................................................................................................. 11

1 資料前處理 ....................................................................................................... 13

1.1 缺值處理 ....................................................................................................... 13

2 機器學習 ........................................................................................................... 14

2.1 邏輯斯迴歸 (Logistic regression)................................................................. 15

2.2 隨機森林 (Random forest) ........................................................................... 15

2.3 極限梯度提升法 (XGBoost) ........................................................................ 15

3 超參數調優 ....................................................................................................... 16

3.1 網格搜尋法 (Grid search)............................................................................. 16

3.2 座標下降法 (Gradient descent) .................................................................... 17

3.3 隨機搜尋法 (Random search) ...................................................................... 17

3.4 貝葉斯搜尋法 (Bayesian optimization) ....................................................... 17

4 模型驗證與評估 ............................................................................................... 18

4.1 交叉驗證 ....................................................................................................... 18

4.2 模型評價指標 ............................................................................................... 19

4.3 生存分析 ....................................................................................................... 21

第四章 實驗設計 .............................................................................................................. 23

V 4.1 資料來源與特徵說明 ....................................................................................... 23

2 資料預處理 ....................................................................................................... 26

3 特徵抽取 ........................................................................................................... 30

3.1 添加自定義字庫 ........................................................................................... 33

3.2 資料切割 ....................................................................................................... 35

4 模型訓練 ........................................................................................................... 36

5 模型驗證 ........................................................................................................... 38

第五章 實驗結果 .............................................................................................................. 40

1 研究樣本特徵 ................................................................................................... 40

2 特徵新增 ........................................................................................................... 41

3 模型訓練 ........................................................................................................... 44

4 模型驗證 ........................................................................................................... 46

5 觀察項目 ........................................................................................................... 49

第六章 結論與未來展望 .................................................................................................. 52 參考資料 ........................................................................................................................... 54

附錄 ................................................................................................................................... 58

Ahn, J. Y., Ryoo, H. W., Park, J., Kim, J. K., Lee, M. J., Kim, J. Y., . . . Kim, Y. A. (2016). New Intervention Model of Regional Transfer Network System to Alleviate Crowding of Regional Emergency Medical Center. J Korean Med Sci, 31(5), 806-

813. doi:10.3346/jkms.2016.31.5.806 Arya, R., Wei, G., McCoy, J. V., Crane, J., Ohman-Strickland, P., & Eisenstein, R. M. (2013). Decreasing length of stay in the emergency department with a split emergency severity index 3 patient flow model. Acad Emerg Med, 20(11), 11711179. doi:10.1111/acem.12249 Asplin, B. R., & Magid, D. J. (2007). If you want to fix crowding, start by fixing your hospital. Ann Emerg Med, 49(3), 273-274. doi:10.1016/j.annemergmed.2007.01.012 Bergstra, J. B., Y. . (2012). Random Search for Hyper-Parameter Optimization. Journal of Machine Learning Research, 13, 25. Brian J. Moore, Carol Stocks, & Owens, P. L. Trends in Emergency Department Visits, 2006-2014 - HCUP STATISTICAL BRIEF #227. Retrieved from https://www.hcup-us.ahrq.gov/reports/statbriefs/sb227-Emergency-DepartmentVisit-Trends.pdf Chalfin, D. B., Trzeciak, S., Likourezos, A., Baumann, B. M., Dellinger, R. P., & group, D.-E. s. (2007). Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med, 35(6), 1477-1483. doi:10.1097/01.CCM.0000266585.74905.5A Chan, H., Lo, S., Lee, L., Lo, W., Yu, W., Wu, Y., . . . Chan, J. (2014). Lean techniques for the improvement of patients' flow in emergency department. World J Emerg Med, 5(1), 24-28. doi:10.5847/wjem.j.1920-8642.2014.01.004

10.5847/wjem.j.issn.1920-8642.2014.01.004 Chen, C. H., Hsieh, J. G., Cheng, S. L., Lin, Y. L., Lin, P. H., & Jeng, J. H. (2020).

Emergency department disposition prediction using a deep neural network with integrated clinical narratives and structured data. Int J Med Inform, 139, 104146. doi:10.1016/j.ijmedinf.2020.104146 Considine, J., Kropman, M., Kelly, E., & Winter, C. (2008). Effect of emergency department fast track on emergency department length of stay: a case-control study. Emerg Med J, 25(12), 815-819. doi:10.1136/emj.2008.057919 Cooke, M. W., Wilson, S., & Pearson, S. (2002). The effect of a separate stream for minor injuries on accident and emergency department waiting times. Emerg Med J, 19(1), 28-30. doi:10.1136/emj.19.1.28

54 Dugas, A. F., Kirsch, T. D., Toerper, M., Korley, F., Yenokyan, G., France, D., . . . Levin,

S. (2016). An Electronic Emergency Triage System to Improve Patient Distribution by Critical Outcomes. J Emerg Med, 50(6), 910-918. doi:10.1016/j.jemermed.2016.02.026 Duguay, C., & Chetouane, F. (2016). Modeling and Improving Emergency Department Systems using Discrete Event Simulation. Simulation, 83(4), 311-320. doi:10.1177/0037549707083111 Gaieski, D. F., Agarwal, A. K., Mikkelsen, M. E., Drumheller, B., Cham Sante, S., Shofer,

F. S., . . . Pines, J. M. (2017). The impact of ED crowding on early interventions and mortality in patients with severe sepsis. Am J Emerg Med, 35(7), 953-960. doi:10.1016/j.ajem.2017.01.061 Handel, D., Epstein, S., Khare, R., Abernethy, D., Klauer, K., Pilgrim, R., . . . Sayan, O.

(2011). Interventions to improve the timeliness of emergency care. Acad Emerg Med, 18(12), 1295-1302. doi:10.1111/j.1553-2712.2011.01230.x HCUPnet. Retrieved from https://hcupnet.ahrq.gov/ Hong, W. S., Haimovich, A. D., & Taylor, R. A. (2018). Predicting hospital admission at emergency department triage using machine learning. PLOS ONE, 13(7), e0201016. doi:10.1371/journal.pone.0201016 Lee, S. Y., Chinnam, R. B., Dalkiran, E., Krupp, S., & Nauss, M. (2020). Prediction of emergency department patient disposition decision for proactive resource allocation for admission. Health Care Manag Sci, 23(3), 339-359. doi:10.1007/s10729-01909496-y Levin, S., Toerper, M., Hamrock, E., Hinson, J. S., Barnes, S., Gardner, H., . . . Kelen, G.

(2018). Machine-Learning-Based Electronic Triage More Accurately Differentiates Patients With Respect to Clinical Outcomes Compared With the Emergency Severity Index. Ann Emerg Med, 71(5), 565-574 e562. doi:10.1016/j.annemergmed.2017.08.005 Mc, D. T. D., Bennett, D. M., & Cameron, P. A. (2004). A paradigm shift in the nature of care provision in emergency departments. Emerg Med J, 21(6), 681-684. doi:10.1136/emj.2004.017640 Moskop, J. C., Sklar, D. P., Geiderman, J. M., Schears, R. M., & Bookman, K. J. (2009).

Emergency department crowding, part 2--barriers to reform and strategies to overcome them. Ann Emerg Med, 53(5), 612-617. doi:10.1016/j.annemergmed.2008.09.024 Mowbray, F., Zargoush, M., Jones, A., de Wit, K., & Costa, A. (2020). Predicting hospital admission for older emergency department patients: Insights from machine learning. Int J Med Inform, 140, 104163. doi:10.1016/j.ijmedinf.2020.104163

55 Naemi, A., Schmidt, T., Mansourvar, M., Naghavi-Behzad, M., Ebrahimi, A., & Wiil, U. K. (2021). Machine learning techniques for mortality prediction in emergency departments: a systematic review. BMJ Open, 11(11), e052663. doi:10.1136/bmjopen-2021-052663 Ng, D., Vail, G., Thomas, S., & Schmidt, N. (2010). Applying the Lean principles of the Toyota Production System to reduce wait times in the emergency department. CJEM, 12(1), 50-57. doi:10.1017/s1481803500012021 Pines, J. M., & McCarthy, M. L. (2011). Executive summary: interventions to improve quality in the crowded emergency department. Acad Emerg Med, 18(12), 12291233. doi:10.1111/j.1553-2712.2011.01228.x Raita, Y., Goto, T., Faridi, M. K., Brown, D. F. M., Camargo, C. A., Jr., & Hasegawa, K. (2019). Emergency department triage prediction of clinical outcomes using machine learning models. Crit Care, 23(1), 64. doi:10.1186/s13054-019-2351-7 Sayan Putatunda, K. R. (2018). A Comparative Analysis of Hyperopt as Against Other Approaches for Hyper-Parameter Optimization of XGBoost. 6–10. doi:https://doi.org/10.1145/3297067.3297080 Spaite, D. W., Bartholomeaux, F., Guisto, J., Lindberg, E., Hull, B., Eyherabide, A., . . . Conroy, C. (2002). Rapid process redesign in a university-based emergency department: decreasing waiting time intervals and improving patient satisfaction. Ann Emerg Med, 39(2), 168-177. doi:10.1067/mem.2002.121215 Sterling, N. W., Patzer, R. E., Di, M., & Schrager, J. D. (2019). Prediction of emergency department patient disposition based on natural language processing of triage notes. Int J Med Inform, 129, 184-188. doi:10.1016/j.ijmedinf.2019.06.008 Sun, Y., Heng, B. H., Tay, S. Y., & Seow, E. (2011). Predicting hospital admissions at emergency department triage using routine administrative data. Acad Emerg Med, 18(8), 844-850. doi:10.1111/j.1553-2712.2011.01125.x Tiffin, A. (2016). Seeing in the Dark: A Machine-Learning Approach to Nowcasting in Lebanon. IMF Working Papers, 16(56). doi:10.5089/9781513568089.001 Tsai, D. J., Tsai, S. H., Chiang, H. H., Lee, C. C., & Chen, S. J. (2022). Development and Validation of an Artificial Intelligence Electrocardiogram Recommendation System in the Emergency Department. J Pers Med, 12(5). doi:10.3390/jpm12050700 van Veelen, M. J., van den Brand, C. L., Reijnen, R., & van der Linden, M. C. (2016). Effects of a general practitioner cooperative co-located with an emergency department on patient throughput. World J Emerg Med, 7(4), 270-273. doi:10.5847/wjem.j.1920-8642.2016.04.005 Ward, M. J., Farley, H., Khare, R. K., Kulstad, E., Mutter, R. L., Shesser, R., & StoneGriffith, S. (2011). Achieving efficiency in crowded emergency departments: a

56 research agenda. Acad Emerg Med, 18(12), 1303-1312. doi:10.1111/j.15532712.2011.01222.x 中華民國衛生福利部 . 109 年醫事機構服務量統計電子書 . Retrieved from https://dep.mohw.gov.tw/DOS/lp-5099-113-xCat-y109.html 中華民國衛生福利部疾病管制署. 醫療照護機構感染管制相關指引. Retrieved from https://www.cdc.gov.tw/Category/MPage/I92jtldmxZO_oolFPzP9HQ 中華民國衛生福利部醫事司. 公告修正急診五級檢傷分類基準. Retrieved from https://dep.mohw.gov.tw/DOMA/lp-5037-106.html 中華民國衛生福利部醫事司. 因應疫情發展醫院應加強人員管制措施. Retrieved from https://www.mohw.gov.tw/cp-4635-51709-1.html 中華民國衛生福利部醫事司 . 急診就醫流程 . Retrieved from https://dep.mohw.gov.tw/doma/cp-3761-39079-106.html

簡易檢索 / 詳目顯示

相關論文