脑卒中多分类预后预测的深度集成优化方法

doi:10.3778/j.issn.1002-8331.2211-0077

摘要/Abstract

摘要： 诊疗前预测急性缺血性脑卒中（AIS）的预后分级，有利于揭示预后转归水平并指导治疗策略，提升方法的预测性能是实现精准医疗的重要指导。利用临床和影像组学的融合特征实施脑卒中的多分类预测，并提出了一种基于融合特征的深度集成优化方法（IABC-DEL）模型，其特征选择方法为Embedded嵌入法和卡方检验，数据不平衡处理方式为Borderline-SMOTE算法，利用Stacking构建深度集成优化模型，基学习器包括深度神经网络（DNN）、长短期记忆网络（LSTM）和门控循环单元（GRU），优化方法为改良人工蜂群算法（IABC）。研究结果表明，深度集成优化方法的预后预测性能优于经典方法和既往研究，Macro-F1 score为87.88%，Macro-AUC为96.27%，ACC为88.02%。因此，基于深度集成优化学习的急性缺血性脑卒中预后模型可对临床诊治和预后康复提供指导意义，并为研究预测提供新的建模思路。

关键词: 深度学习, 集成学习, 人工蜂群算法, 急性缺血性脑卒中, 影像组学

Abstract: The prognosis grading of acute ischemic stroke（AIS） can effectively assess the prognosis of patients and guide clinical treatment. Improving the method’s prediction performance is an important step toward precision medicine. This paper uses clinical and radiomics fusion features to implement multiclass stroke prediction and proposes a deep integration optimization method. Embedded embedding and the chi-square test are two methods for selecting features. To deal with data imbalance, the Borderline-SMOTE algorithm is used. Stacking is employed in the construction of a deep integration optimization model. Deep neural networks（DNN）, long-short term memory（LSTM）, and gated recurrent unit（GRU） are basic learners. The improved artificial bee colony（IABC） algorithm is used for optimization. The research findings show that the deep integration optimization method outperforms the classical method and advanced research in terms of prognosis prediction performance. Macro-F1 score is 87.88%, Macro-AUC is 96.27%, and ACC is 88.02%. As a result, the acute ischemic stroke prognosis model based on deep integration optimization learning can provide guidance for clinical diagnosis, treatment, and prognosis rehabilitation, as well as new modeling ideas for research and prediction.

Key words: deep learning, integrated learning, artificial bee colony algorithm, acute ischemic stroke, radiomics

叶伟, 陶永军, 陈锡程, 伍亚舟. 脑卒中多分类预后预测的深度集成优化方法[J]. 计算机工程与应用, 2023, 59(5): 95-105.

YE Wei, TAO Yongjun, CHEN Xicheng, WU Yazhou. Deep Ensemble Evolutionary Multi-Classification Method for Predicting Prognosis of Stroke[J]. Computer Engineering and Applications, 2023, 59(5): 95-105.

参考文献

[1] HERPICH F，RINCON F.Management of acute ischemic stroke[J].Crit Care Med，2020，48（11）：1654-1663.
[2] DING Q，LIU S，YAO Y，et al.Global，regional，and national burden of ischemic stroke，1990—2019[J].Neurology，2022，98（3）：e279-e290.?
[3] GBD 2019 Stroke Collaborators.Global，regional，and national burden of stroke and its risk factors，1990—2019：a systematic analysis for the Global Burden of Disease Study 2019[J].Lancet Neurol，2021，20（10）：795-820.?
[4] MATSUMOTO K，NOHARA Y，SOEJIMA H，et al.Stroke prognostic scores and data-driven prediction of clinical outcomes after acute ischemic stroke[J].Stroke，2020，51（5）：1477-1483.
[5] AVERY E W，BEHLAND J，MAK A，et al.Dataset on acute stroke risk stratification from CT angiographic radiomics[J].Data Brief，2022，44（1）：108542.
[6] SUNG S F，HSIEH C Y，HU Y H.Early prediction of functional outcomes after acute ischemic stroke using unstructured clinical text：retrospective cohort study[J].JMIR Med Inform，2022，10（2）：e29806.
[7] WANG H，SUN Y，GE Y，et al.A clinical-radiomics nomogram for functional outcome predictions in ischemic stroke[J].Neurology and Therapy，2021，10（2）：819-832.
[8] MONTELLANO F A，UNGETHUM K，RAMIRO L，et al.Role of blood-based biomarkers in ischemic stroke prognosis：a systematic review[J].Stroke，2021，52（2）：543-551.
[9] HEO J，YOON J G，PARK H，et al.Machine learning-based model for prediction of outcomes in acute stroke[J].Stroke，2019，50（5）：1263-1265.
[10] HWANGBO L，KANG Y J，KWON H，et al.Stacking ensemble learning model to predict 6-month mortality in ischemic stroke patients[J].Sci Rep，2022，12（1）：17389.
[11] WU H，ZHANG H，KARWATH A，et al.Ensemble learning for poor prognosis predictions：a case study on SARS-CoV-2[J].J Am Med Inform Assoc，2021，28（4）：791-800.
[12] PARK D，CHO J M，YANG J W，et al.Classification of expert-level therapeutic decisions for degenerative cervical myelopathy using ensemble machine learning algorithms[J].Front Surg，2022，9（1）：1010420.
[13] LERTAMPAIPORN S，HONGSTHONG A，WATTANAPORNPROM W，et al.Ensemble-AHTPpred：a robust ensemble machine learning model integrated with a new composite feature for identifying antihypertensive peptides[J].Front Genet，2022，13（1）：883766.
[14] SHARMA D，GOTLIEB N，FARKOUH M E，et al.Machine learning approach to classify cardiovascular disease in patients with nonalcoholic fatty liver disease in the UK biobank cohort[J].J Am Heart Assoc，2022，11（1）：e22576.
[15] SHI Y，GUO L，CHEN Y，et al.Risk factors for ischemic stroke：differences between cerebral small vessel and large artery atherosclerosis aetiologies[J].Folia Neuropathol，2021，59（4）：378-385.
[16] QURESHI A I，BASKETT W I，HUANG W，et al.Acute Ischemic stroke and COVID-19：an analysis of 27?676 Patients[J].Stroke，2021，52（3）：905-912.
[17] MILLER R，KLYDE D M，UNDA S R，et al.Classical risk factors for ischemic stroke are not associated with inpatient post-stroke mortality in sickle cell disease[J].J Stroke Cerebrovasc Dis，2021，30（12）：106089.
[18] ELSAID N，BIGLIARDI G，DELL’ACQUA M L，et al.Evaluation of stroke prognostication using age and NIH Stroke Scale index（SPAN-100 index） in delayed intravenous thrombolysis patients（beyond 4.5 hours）[J].J Stroke Cerebrovasc Dis，2022，31（4）：106384.
[19] LARSEN K，JAEGER H S，HOV M R，et al.Streamlining acute stroke care by introducing national institutes of health stroke scale in the emergency medical services：a prospective cohort study[J].Stroke，2022，53（6）：2050-2057.
[20] VAN GRIETHUYSEN J，FEDOROV A，PARMAR C，et al.Computational radiomics system to decode the radiographic phenotype[J].Cancer Res，2017，77（21）：e104-e107.
[21] NING Q，ZHAO X，MA Z.A novel method for identification of glutarylation sites combining borderline-SMOTE with tomek links technique in imbalanced data[J].IEEE/ACM Trans Comput Biol Bioinform，2022，19（5）：2632-2641.
[22] HOCHREITER S，SCHMIDHUBER J.Long short-term memory[J].Neural Computation，1997，9（8）：1735-1780.
[23] GERS F A，SCHMIDHUBER J，CUMMINS F.Learning to forget：continual prediction with LSTM[J].Neural Computation，2000，12（10）：2451-2471.
[24] MUNIR H S，REN S，MUSTAFA M，et al.Attention based GRU-LSTM for software defect prediction[J].PLoS One，2021，16（3）：e0247444.
[25] MORGAN-BENITA J A，GALVáN-TEJADA C E，CRUZ M，et al.Hard voting ensemble approach for the detection of type 2 diabetes in mexican population with non-glucose related features[J].Healthcare（Basel），2022，10（8）：1362.?
[26] ALI S，HUSSAIN A，AICH S，et al.A soft voting ensemble-based model for the early prediction of idiopathic pulmonary fibrosis（IPF） disease severity in lungs disease patients[J].Life（Basel），2021，11（10）：1092.
[27] MANCONI A，ARMANO G，GNOCCHI M，et al.A soft-voting ensemble classifier for detecting patients affected by COVID-19[J].Applied Sciences，2022，12（15）：7554.
[28] WANG Y，PAN Q，LIU X，et al.ET-MSF：a model stacking framework to identify electron transport proteins[J].Front Biosci（Landmark Ed），2022，27（1）：12.
[29] CHATTERJEE S，BYUN Y C.EEG-based emotion classification using stacking ensemble approach[J].Sensors（Basel），2022，22（21）：8550.
[30] ALAKA S A，MENON B K，BROBBEY A，et al.Functional outcome prediction in ischemic stroke：a comparison of machine learning algorithms and regression models[J].Front Neurol，2020，11（1）：889.
[31] TOLHUISEN M L，HOVING J W，KOOPMAN M S，et al.Outcome prediction based on automatically extracted infarct core image features in patients with acute ischemic stroke[J].Diagnostics（Basel），2022，12（8）：1786.
[32] HOFMEISTER J，BERNAVA G，ROSI A，et al.Clot-based radiomics predict a mechanical thrombectomy strategy for successful recanalization in acute ischemic stroke[J].Stroke，2020，51（1）：2488-2494.