運用機器學習進行透析中低血壓之偵測

Detecting Intradialytic Hypotension Using Machine Learning

陳欣惠(Xin-Hui Tan)

指導教授 ： 張智星

國立臺灣大學/電機資訊學院/資訊工程學系/碩士(2023年)

https://doi.org/10.6342/NTU202302763

摘要

慢性腎臟病是台灣重要的醫療課題，約九成的慢性腎臟病末期患者會採取腎臟替代療法中的血液透析進行治療。而透析中低血壓(intradialytic hypotension, IDH)是血液透析患者常見的併發症，有研究表明 IDH 會增加患者心血管疾病發病率和死亡率。本研究合理運用台大醫院血液透析患者的透析數據，採用較為嚴格之 IDH 定義(考量患者發生的症狀以及醫護人員的干預和處置)對資料集進行標注，有別於其他研究工作僅以透析中的血壓變化來定義 IDH。基於台大醫院的透析資料集，設計若干實驗以探討機器學習在即時偵測和預測未來有症狀之 IDH 的可行性。實驗結果顯示，XGBoost 和 RF 演算法在即時偵測有症狀之 IDH 的靈敏度和特異度皆可達 0.8;LSTM 和 MLP 在預測未來有症狀之 IDH 時，擴增輸入的時間序列資料，模型預測的能力得以提升。

關鍵字

透析中低血壓 ； 慢性腎臟病 ； 機器學習 ； 深度學習 ； 類別不平衡 ； 時間序列分類

並列摘要

Chronic Kidney Disease (CKD) is an important medical issue in Taiwan. Approximately 90% of end-stage renal disease (ESRD) patients are treated with hemodialysis. Intradialytic hypotension (IDH) is a common complication among hemodialysis patients. Studies have shown that IDH increases the incidence of cardiovascular diseases and mortality. Considering the intradialytic symptoms and the intradialytic interventions, with the hemodialysis data from National Taiwan University Hospital, this thesis aims to employ a rigorous definition of IDH to define IDH during dialysis treatment, unlike other studies that solely relies on blood pressure drops. Our experiments are designed to explore the feasibility of Machine Learning in detection and prediction of IDH with symptoms. The experimental results demonstrate that XGBoost and Random Forest algorithms achieve a sensitivity and specificity of 0.8 in the detection of IDH with symptoms. When we provide more time series data, the performance of LSTM and MLP models for predicting IDH with symptoms are improved.

並列關鍵字

Intradialytic Hypotension ； Chronic Kidney Disease ； Machine Learning ； Deep Learning ； Class Imbalance ； Time Series Classification

參考文獻

M. M. Ahsan, S. A. Luna, and Z. Siddique. Machine-learning-based disease diagnosis: A comprehensive review. Healthcare, 10(3):541, Mar 2022.


Z. Alhassan, A. S. McGough, R. Alshammari, T. Daghstani, D. Budgen, and N. Al Moubayed. Type-2 diabetes mellitus diagnosis from time series clinical data using deep learning models. In V. Kůrková, Y. Manolopoulos, B. Hammer, L. Iliadis, and I. Maglogiannis, editors, Artificial Neural Networks and Machine Learning – ICANN 2018, pages 468–478, Cham, 2018. Springer International Publishing.


M. M. Assimon and J. E. Flythe. Definitions of intradialytic hypotension. Seminars in Dialysis, 30(6):pp. 464–472, 2017.


B. E. Boser, I. M. Guyon, and V. N. Vapnik. A training algorithm for optimal margin classifiers. In Proceedings of the Fifth Annual Workshop on Computational Learning Theory, COLT ’92, page 144–152, New York, NY, USA, 1992. Association for Computing Machinery.


L. Bottou. Stochastic gradient learning in neural networks. Proceedings of Neuro-Nımes, 91(8):12, 1991.