臺灣博碩士論文加值系統

近年來，機器學習和深度學習運用電子醫療系統的資料，有效地預測醫療中的風險發生。在臺灣，我們有每位病人的術前麻醉評估資料和整個醫療過程的電子病歷紀錄。在日新月異的機器學習和自然語言處理發展下，這些資料可用作輸入特徵訓練模型來更準確地預測風險。術前診斷和手術方式的文字描述可以幫助麻醉科醫師了解病人手術相關風險，但運用自然語言處理進行術式和診斷的文字的風險預測還有待研究。因此，本計畫預計使用患者的圍手術期的資料，使用機器學習和自然語言處理建立模型，預測病人的死亡率。
本回溯性研究收案包含所有接受全身或脊椎麻醉之受試者，取得術前資料作為輸入的特徵，包括病人基本資料、共病症、實驗室檢查資料以及診斷和手術方式的文字資料，訓練模型學習是否三十天內院內死亡。資料類型包括類別變項、連續變項和文字資料，運用機器學習（random forest, eXtreme Gradient Boost, logistic regression）、深度學習和自然語言處理進行訓練。先比較自然語言處理中Bidirectional Encoder Representations from Transformers (BERT)、Embeddings from Language Models 和 Global Vector等方法，表現最好的方法再用來與deep neural network (DNN) 結合成混成模型。先比較機器學習和DNN用一樣的輸入特徵的表現，接著計算 area under the receiver operating characteristic curve (AUROC) 和area under the precision-recall curve (AUPRC)比較加入文字資料的效果。
在語言模型的比較部分，運用術前診斷和手術名稱預測術後死亡時，bio-clinical BERT模型相較於其他語言模型有更高的AUROC (0.883)。在比較合併結構化與非結構化資料的模型部分，BERT-DNN模型有最高的AUROC (0.964) 和 AUPRC (0.336)。BERT-DNN的AUROC顯著性地高於eXtreme Gradient Boost classifier、logistic regression和American Society of Anesthesiologists physical status，但不顯著性高於DNN和random forest classifier。BERT-DNN的AUPRC顯著性地高於其他模型。
描述手術的文字對於預測術後死亡率很重要。本研究將術前診斷和手術方法運用BERT語言模型轉成向量，並且與深度神經網路結合用以預測術後死亡率。這個預測模型能夠用結構化資料和文字辨識高風險病患族群，可以減少遺漏，並且及早介入處置、溝通和管理醫療資源。

Using machine learning techniques has resulted in more accurate predictions of postoperative mortality compared to previous methods. Before a patient undergoes a surgery, we had free text descriptions of the preoperative diagnosis and the planned procedure. Since reading these descriptions can assist anesthesiologists in assessing the risk of the surgery, we hypothesized that deep learning models utilizing unstructured text can enhance postoperative mortality prediction. However, it can be difficult to extract useful concept embeddings from unstructured clinical text.
The goal of this study is to develop a deep learning model that combines structured and unstructured features in order to predict the risk of 30-day postoperative mortality within a hospital before surgery. The effectiveness of machine learning models that use preoperative data, with or without free clinical text, in predicting postoperative mortality will be evaluated.
In this study, we retrospectively gathered discharge summaries, surgical information, and preoperative anesthesia assessment of patients who received general or neuraxial anesthesia from electronic medical records. We first compared different natural language processing methods, including Global Vector, Embeddings from Language Models, and Bidirectional Encoder Representations from Transformers (BERT). Then, we combined the top-performing aforementioned method with a deep neural network (DNN) model to extract information from clinical texts. We compared the performance of machine learning models, including random forest, eXtreme gradient boost, and logistic regression, to the DNN model using the same input features. We assessed the impact of adding text information on model performance by measuring the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). Statistical significance was determined using P < .05.
In comparing performance of embedding methods, the bio-clinical BERT model had the highest AUROC of 0.883 to predict postoperative mortality with text of preoperative diagnosis and procedures among all language models. In comparing performance of combined structured and unstructured models, BERT-DNN had the highest AUROC of 0.964 (95% confidence interval [CI]: 0.961 – 0.967) and the highest AUPRC of 0.336 (95% CI: 0.276 – 0.402). The BERT-DNN was significantly higher than the eXtreme gradient boost classifier, logistic regression, and American Society of Anesthesiologists physical status in AUROC, but not significantly higher than the DNN and random forest classifier. The BERT-DNN was significantly higher than other models in AUPRC.
In summary, the surgical text descriptions were crucial for predicting postoperative mortality. In this study, we used deep learning models to combine the word embeddings of preoperative diagnoses and planned procedures, obtained using the contextualized language model BERT, to predict postoperative mortality. This ability to predict risk can help identify patients who are at higher risk based on the structured data and text in electronic health records, potentially reducing missed opportunities for early intervention, communication, and resource management.

口試委員會審定書 i
誌謝 ii
中文摘要 iii
ABSTRACT v
CONTENTS vii
LIST OF FIGURES x
LIST OF TABLES xii
Chapter 1 Introduction 1
1.1 Anesthesia and Perioperative Medicine 1
1.2 Anesthesia Risk Prediction 2
1.3 Surgery Categories and Risk Stratification 4
1.4 Aim 5
Chapter 2 Literature Review 6
2.1 Conventional Tools Predicting Postoperative Mortality 6
2.2 Machine Learning Method Predicting Postoperative Mortality 10
2.3 Unstructured Free Text for Risk Stratification 14
Chapter 3 Method 17
3.1 Data Extraction 17
3.2 Data Description 18
3.3 Input Features 23
3.3.1 Structured data 25
3.3.2 Unstructured free text 27
3.4 Comparing Different Embedding Methods for Stratifying Surgery Risk 27
3.5 BERT Model Stratifying Preoperative Text 28
3.6 Comparing Different Structures of BERT-DNN Model 29
3.7 Comparing the Model with and without Input Free Text 32
3.8 Model Evaluation 34
3.9 Model of a Reduced Set of Features 37
3.10 Word Embeddings Visualization 38
3.11 Feature Importance 38
3.12 Package Version and Statistical Software 39
Chapter 4 Results 40
4.1 Comparing Different Language Models for Stratifying Surgery Risk 40
4.2 Comparing Model Performance of Different Structures 40
4.3 Comparing Machine Learning Models Performance 42
4.4 Statistical Significance 49
4.5 Calibration Plot 51
4.6 Visualization of Word Embeddings 52
4.7 Cases Comparison between BERT-DNN and DNN 57
4.8 Comparing Model Performances with Reduced Set of Features 60
4.9 SHAP 65
Chapter 5 Discussion 70
5.1 Principal Findings 70
5.2 Comparing Results with Relative Studies 70
5.3 Effect of Text on Model Prediction 74
5.4 Implement of the Prediction Model in Electronic Health System 75
5.5 Limitations 77
Chapter 6 Conclusion 80
6.1 Future Work 81
REFERENCES 83
ABBREVIATION 94

[1]A.G. Hudetz, General Anesthesia and Human Brain Connectivity, Brain Connect. 2 (2012) 291–302. https://doi.org/10.1089/brain.2012.0107.
[2]A.F. Merry, S.J. Mitchell, Complications of anaesthesia, Anaesthesia. 73 (2018) 7–11. https://doi.org/10.1111/anae.14135.
[3]A.S. Patel, A. Bergman, B.W. Moore, U. Haglund, The Economic Burden of Complications Occurring in Major Surgical Procedures: a Systematic Review, Appl Health Econ Health Policy. 11 (2013) 577–592. https://doi.org/10.1007/s40258-013-0060-y.
[4]M.B. Ferschl, A. Tung, B. Sweitzer, D. Huo, D.B. Glick, Preoperative Clinic Visits Reduce Operating Room Cancellations and Delays, Anesthesiology. 103 (2005) 855–859. https://doi.org/10.1097/00000542-200510000-00025.
[5]R.M. Pearse, P.A. Clavien, N. Demartines, L.A. Fleisher, M. Grocott, et al., Global patient outcomes after elective surgery: Prospective cohort study in 27 low-, middle- and high-income countries, Br J Anaesth. 117 (2016) 601–609. https://doi.org/10.1093/bja/aew316.
[6]M.P.W. Grocott, R.M. Pearse, Perioperative medicine: the future of anaesthesia?, (2012).
[7]K.Y. Bilimoria, Y. Liu, J.L. Paruch, L. Zhou, T.E. Kmiecik, et al., Development and evaluation of the universal ACS NSQIP surgical risk calculator: a decision aid and informed consent tool for patients and surgeons, J Am Coll Surg. 217 (2013) 833–842.
[8]Y. le Manach, G. Collins, R. Rodseth, C. le Bihan-Benjamin, B. Biccard, et al., Preoperative Score to Predict Postoperative Mortality (POSPOM): Derivation and validation, Anesthesiology. 124 (2016) 570–579. https://doi.org/10.1097/ALN.0000000000000972.
[9]J.E. Dalton, A. Kurz, A. Turan, E.J. Mascha, D.I. Sessler, et al., Development and Validation of a Risk Quantification Index for 30-Day Postoperative Mortality and Morbidity in Noncardiac Surgical Patients, Anesthesiology. 114 (2011) 1336–1344. https://doi.org/10.1097/ALN.0b013e318219d5f9.
[10]D.I. Sessler, J.C. Sigl, P.J. Manberg, S.D. Kelley, A. Schubert, et al., Broadly Applicable Risk Stratification System for Predicting Duration of Hospitalization and Mortality, Anesthesiology. 113 (2010) 1026–1037. https://doi.org/10.1097/ALN.0b013e3181f79a8d.
[11]B.L. Hill, R. Brown, E. Gabel, N. Rakocz, C. Lee, et al., An automated machine learning-based model predicts postoperative mortality using readily-extractable preoperative electronic health record data, Br J Anaesth. 123 (2019) 877–886. https://doi.org/10.1016/j.bja.2019.07.030.
[12]C.K. Lee, I. Hofer, E. Gabel, P. Baldi, M. Cannesson, Development and Validation of a Deep Neural Network Model for Prediction of Postoperative In-hospital Mortality, Anesthesiology. 129 (2018) 649–662. https://doi.org/10.1097/ALN.0000000000002186.
[13]B.A. Fritz, Z. Cui, M. Zhang, Y. He, Y. Chen, et al., Deep-learning model for predicting 30-day postoperative mortality, Br J Anaesth. 123 (2019) 688–695. https://doi.org/10.1016/j.bja.2019.07.025.
[14]J. Bucerius, J.F. Gummert, M.A. Borger, T. Walther, N. Doll, et al., Stroke after cardiac surgery: a risk factor analysis of 16,184 consecutive adult patients, Ann Thorac Surg. 75 (2003) 472–478. https://doi.org/10.1016/S0003-4975(02)04370-9.
[15]K.B. Kaufmann, T. Loop, S. Heinrich, Risk factors for post‐operative pulmonary complications in lung cancer patients after video‐assisted thoracoscopic lung resection: Results of the German Thorax Registry, Acta Anaesthesiol Scand. 63 (2019) 1009–1018. https://doi.org/10.1111/aas.13388.
[16]G.E. Weissman, R.A. Hubbard, L.H. Ungar, M.O. Harhay, C.S. Greene, et al., Inclusion of Unstructured Clinical Text Improves Early Prediction of Death or Prolonged ICU Stay, Crit Care Med. 46 (2018) 1125—1132. https://doi.org/10.1097/ccm.0000000000003148.
[17]D. Zhang, C. Yin, J. Zeng, X. Yuan, P. Zhang, Combining structured and unstructured data for predictive models: a deep learning approach, BMC Med Inform Decis Mak. 20 (2020) 1–11.
[18]D. Mayhew, V. Mendonca, B.V.S. Murthy, A review of ASA physical status–historical perspectives and modern developments, Anaesthesia. 74 (2019) 373–379.
[19]ASA House of Delegates, ASA physical status classification system, ASA House of Delegates. (2014). https://www.asahq.org/resources/clinical-information/asa-physical-status-classification-system (accessed June 5, 2022).
[20]S.C. FARROW, F.G.R. FOWKES, J.N. LUNN, I.B. ROBERTSON, P. SAMUEL, EPIDEMIOLOGY IN ANAESTHESIA II: FACTORS AFFECTING MORTALITY IN HOSPITAL, Br J Anaesth. 54 (1982) 811–817. https://doi.org/10.1093/bja/54.8.811.
[21]A. Sankar, S.R. Johnson, W.S. Beattie, G. Tait, D.N. Wijeysundera, Reliability of the American Society of Anesthesiologists physical status scale in clinical practice, Br J Anaesth. 113 (2014) 424–432. https://doi.org/10.1093/bja/aeu100.
[22]B. Horvath, B. Kloesel, M.M. Todd, D.J. Cole, R.C. Prielipp, The Evolution, Current Value, and Future of the American Society of Anesthesiologists Physical Status Classification System, Anesthesiology. 135 (2021) 904–919. https://doi.org/10.1097/ALN.0000000000003947.
[23]E.E. Hurwitz, M. Simon, S.R. Vinta, C.F. Zehm, S.M. Shabot, et al., Adding examples to the ASA-physical status classification improves correct assignment to patients, Anesthesiology. 126 (2017) 614–622.
[24]C. Curatolo, A. Goldberg, D. Maerz, H.M. Lin, H. Shah, et al., ASA physical status assignment by non-anesthesia providers: Do surgeons consistently downgrade the ASA score preoperatively?, J Clin Anesth. 38 (2017) 123–128. https://doi.org/10.1016/j.jclinane.2017.02.002.
[25]M.J.G. Sigakis, E.A. Bittner, J.P. Wanderer, Validation of a risk stratification index and risk quantification index for predicting patient outcomes: in-hospital mortality, 30-day mortality, 1-year mortality, and length-of-stay, Anesthesiology. 119 (2013) 525–540.
[26]D.A. Hashimoto, E. Witkowski, L. Gao, O. Meireles, G. Rosman, Artificial intelligence in anesthesiology: current techniques, clinical applications, and limitations, Anesthesiology. 132 (2020) 379–394.
[27]S. Kendale, P. Kulkarni, A.D. Rosenberg, J. Wang, Supervised Machine-learning Predictive Analytics for Prediction of Postinduction Hypotension, Anesthesiology. 129 (2018) 675–688. https://doi.org/10.1097/ALN.0000000000002374.
[28]T. van den Bosch, A.-L.K. Warps, M.P.M. de Nerée tot Babberich, C. Stamm, B.F. Geerts, et al., Predictors of 30-Day Mortality Among Dutch Patients Undergoing Colorectal Cancer Surgery, 2011-2016, JAMA Netw Open. 4 (2021). https://doi.org/10.1001/jamanetworkopen.2021.7737.
[29]H. Zhao, J. You, Y. Peng, Y. Feng, Machine Learning Algorithm Using Electronic Chart-Derived Data to Predict Delirium After Elderly Hip Fracture Surgeries: A Retrospective Case-Control Study, Front Surg. 8 (2021). https://doi.org/10.3389/fsurg.2021.634629.
[30]Y. Wang, L. Lei, M. Ji, J. Tong, C.M. Zhou, et al., Predicting postoperative delirium after microvascular decompression surgery with machine learning, J Clin Anesth. 66 (2020). https://doi.org/10.1016/j.jclinane.2020.109896.
[31]S. Lee, H.C. Lee, Y.S. Chu, S.W. Song, G.J. Ahn, et al., Deep learning models for the prediction of intraoperative hypotension, Br J Anaesth. 126 (2021) 808–817. https://doi.org/10.1016/j.bja.2020.12.035.
[32]S. Datta, T.J. Loftus, M.M. Ruppert, C. Giordano, G.R. Upchurch, et al., Added Value of Intraoperative Data for Predicting Postoperative Complications: The MySurgeryRisk PostOp Extension, Journal of Surgical Research. 254 (2020) 350–363. https://doi.org/10.1016/j.jss.2020.05.007.
[33]J. Ye, L. Yao, J. Shen, R. Janarthanam, Y. Luo, Predicting mortality in critically ill patients with diabetes using machine learning and clinical notes, BMC Med Inform Decis Mak. 20 (2020) 1–7.
[34]A. Bonde, K.M. Varadarajan, N. Bonde, A. Troelsen, O.K. Muratoglu, et al., Assessing the utility of deep neural networks in predicting postoperative surgical complications: a retrospective study, Lancet Digit Health. 3 (2021) e471–e485. https://doi.org/10.1016/S2589-7500(21)00084-4.
[35]X. Yan, J. Goldsmith, S. Mohan, Z.A. Turnbull, R.E. Freundlich, et al., Impact of Intraoperative Data on Risk Prediction for Mortality After Intra-Abdominal Surgery, Anesth Analg. 134 (2021) 102–113.
[36]A.R. Kang, J. Lee, W. Jung, M. Lee, S.Y. Park, et al., Development of a prediction model for hypotension after induction of anesthesia using machine learning, PLoS One. 15 (2020) e0231172.
[37]T. Konishi, T. Goto, M. Fujiogi, N. Michihata, R. Kumazawa, et al., New machine learning scoring system for predicting postoperative mortality in gastroduodenal ulcer perforation: A study using a Japanese nationwide inpatient database, Surgery. (2021). https://doi.org/10.1016/J.SURG.2021.08.031.
[38]A. Rajkomar, E. Oren, K. Chen, A.M. Dai, N. Hajaj, et al., Scalable and accurate deep learning with electronic health records, NPJ Digit Med. 1 (2018) 18. https://doi.org/10.1038/s41746-018-0029-1.
[39]J.B. Bijker, S. Persoon, L.M. Peelen, K.G.M. Moons, C.J. Kalkman, et al., Intraoperative Hypotension and Perioperative Ischemic Stroke after General Surgery, Anesthesiology. 116 (2012) 658–664. https://doi.org/10.1097/ALN.0b013e3182472320.
[40]M. Walsh, P.J. Devereaux, A.X. Garg, A. Kurz, A. Turan, et al., Relationship between Intraoperative Mean Arterial Pressure and Clinical Outcomes after Noncardiac Surgery, Anesthesiology. 119 (2013) 507–515. https://doi.org/10.1097/ALN.0b013e3182a10e26.
[41]E.J. Mascha, D. Yang, S. Weiss, D.I. Sessler, Intraoperative Mean Arterial Pressure Variability and 30-day Mortality in Patients Having Noncardiac Surgery, Anesthesiology. 123 (2015) 79–91. https://doi.org/10.1097/ALN.0000000000000686.
[42]A. Gregory, W.H. Stapelfeldt, A.K. Khanna, N.J. Smischney, I.J. Boero, et al., Intraoperative Hypotension Is Associated with Adverse Clinical Outcomes after Noncardiac Surgery, Anesth Analg. (2021) 1654–1665. https://doi.org/10.1213/ANE.0000000000005250.
[43]B.A. Fritz, M. Abdelhack, C.R. King, Y. Chen, M.S. Avidan, Update to ‘Deep-learning model for predicting 30-day postoperative mortality’ (Br J Anaesth 2019; 123: 688–95), Br J Anaesth. 125 (2020) e230–e231. https://doi.org/10.1016/j.bja.2020.04.010.
[44]T. Mikolov, I. Sutskever, K. Chen, G.S. Corrado, J. Dean, Distributed representations of words and phrases and their compositionality, in: Adv Neural Inf Process Syst, 2013: pp. 3111–3119.
[45]J. Pennington, R. Socher, C.D. Manning, Glove: Global vectors for word representation, in: Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2014: pp. 1532–1543.
[46]M.E. Peters, M. Neumann, M. Iyyer, M. Gardner, C. Clark, et al., Deep contextualized word representations, ArXiv Preprint ArXiv:1802.05365. (2018).
[47]J. Devlin, M.-W. Chang, K. Lee, K. Toutanova, BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding, ArXiv Preprint ArXiv:1810.04805. (2018).
[48]Y. Si, J. Wang, H. Xu, K. Roberts, Enhancing clinical concept extraction with contextual embeddings, Journal of the American Medical Informatics Association. 26 (2019) 1297–1304. https://doi.org/10.1093/jamia/ocz096.
[49]J. Lee, W. Yoon, S. Kim, D. Kim, S. Kim, et al., BioBERT: a pre-trained biomedical language representation model for biomedical text mining, (2019). https://doi.org/10.1093/bioinformatics/btz682.
[50]E. Alsentzer, J.R. Murphy, W. Boag, W.-H. Weng, D. Jin, et al., Publicly available clinical BERT embeddings, ArXiv Preprint ArXiv:1904.03323. (2019).
[51]Y.P. Chen, Y.Y. Chen, J.J. Lin, C.H. Huang, F. Lai, Modified bidirectional encoder representations from transformers extractive summarization model for hospital information systems based on character-level tokens (AlphaBERT): Development and performance evaluation, JMIR Med Inform. 8 (2020). https://doi.org/10.2196/17787.
[52]P.-F. Chen, S.-M. Wang, W.-C. Liao, L.-C. Kuo, K.-C. Chen, et al., Automatic ICD-10 Coding and Training System: Deep Neural Network Based on Supervised Learning, JMIR Med Inform. 9 (2021) e23230. https://doi.org/10.2196/23230.
[53]P.-F. Chen, L. Chen, Y.-K. Lin, G.-H. Li, F. Lai, et al., Predicting Postoperative Mortality With Deep Neural Networks and Natural Language Processing: Model Development and Validation, JMIR Med Inform. 10 (2022) e38241. https://doi.org/10.2196/38241.
[54]E. Loper, S. Bird, Nltk: The natural language toolkit, ArXiv Preprint Cs/0205028. (2002).
[55]N. v Chawla, K.W. Bowyer, L.O. Hall, W.P. Kegelmeyer, SMOTE: synthetic minority over-sampling technique, Journal of Artificial Intelligence Research. 16 (2002) 321–357.
[56]T. Saito, M. Rehmsmeier, The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets, PLoS One. 10 (2015) e0118432.
[57]B. Ozenne, F. Subtil, D. Maucort-Boulch, The precision–recall curve overcame the optimism of the receiver operating characteristic curve in rare diseases, J Clin Epidemiol. 68 (2015) 855–859. https://doi.org/https://doi.org/10.1016/j.jclinepi.2015.02.010.
[58]E.R. DeLong, D.M. DeLong, D.L. Clarke-Pearson, Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach, Biometrics. (1988) 837–845.
[59]K. Boyd, K.H. Eng, C.D. Page, Area under the Precision-Recall Curve: Point Estimates and Confidence Intervals, in: H. Blockeel, K. Kersting, S. Nijssen, F. Železný (Eds.), Machine Learning and Knowledge Discovery in Databases, Springer Berlin Heidelberg, Berlin, Heidelberg, 2013: pp. 451–466.
[60]L. van der Maaten, G. Hinton, Visualizing data using t-SNE., Journal of Machine Learning Research. 9 (2008).
[61]M. Jin, M.T. Bahadori, A. Colak, P. Bhatia, B. Celikkaya, et al., Improving hospital mortality prediction with medical named entities and multimodal learning, ArXiv Preprint ArXiv:1811.12276. (2018).
[62]S.M. Lundberg, S.-I. Lee, A unified approach to interpreting model predictions, in: Proceedings of the 31st International Conference on Neural Information Processing Systems, 2017: pp. 4768–4777.
[63]F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, et al., Scikit-learn: Machine learning in Python, The Journal of Machine Learning Research. 12 (2011) 2825–2830.
[64]A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, et al., Pytorch: An imperative style, high-performance deep learning library, Adv Neural Inf Process Syst. 32 (2019).
[65]J.E. Dalton, A. Kurz, A. Turan, E.J. Mascha, D.I. Sessler, et al., Development and Validation of a Risk Quantification Index for 30-Day Postoperative Mortality and Morbidity in Noncardiac Surgical Patients, Survey of Anesthesiology. 56 (2012) 193.
[66]P.-F. Chen, T.-L. He, S.-C. Lin, Y.-C. Chu, C.-T. Kuo, et al., Training a Deep Contextualized Language Model for International Classification of Diseases, 10th Revision Classification via Federated Learning: Model Development and Validation Study, JMIR Med Inform. 10 (2022) e41342. https://doi.org/10.2196/41342.
[67]P.-F. Chen, K.-C. Chen, W.-C. Liao, F. Lai, T.-L. He, et al., Automatic International Classification of Diseases Coding System: Deep Contextualized Language Model With Rule-Based Approaches, JMIR Med Inform. 10 (2022) e37557. https://doi.org/10.2196/37557.
[68]Y. Liu, M. Ott, N. Goyal, J. Du, M. Joshi, et al., RoBERTa: A Robustly Optimized BERT Pretraining Approach, (2019).
[69]S.C. Huang, A. Pareek, S. Seyyedi, I. Banerjee, M.P. Lungren, Fusion of medical imaging and electronic health records using deep learning: a systematic review and implementation guidelines, NPJ Digit Med. 3 (2020). https://doi.org/10.1038/s41746-020-00341-z.
[70]B.J. Marafino, M. Park, J.M. Davies, R. Thombley, H.S. Luft, et al., Validation of Prediction Models for Critical Care Outcomes Using Natural Language Processing of Electronic Health Record Data, JAMA Netw Open. 1 (2018). https://doi.org/10.1001/jamanetworkopen.2018.5097.
[71]O.J. Bear Don’t Walk IV, T. Sun, A. Perotte, N. Elhadad, Clinically relevant pretraining is all you need, Journal of the American Medical Informatics Association. (2021). https://doi.org/10.1093/jamia/ocab086.
[72]H.-I. Lin, M.C. Nguyen, Boosting minority class prediction on imbalanced point cloud data, Applied Sciences. 10 (2020) 973.
[73]J.M. Johnson, T.M. Khoshgoftaar, Survey on deep learning with class imbalance, J Big Data. 6 (2019) 1–54.
[74]B.K. Beaulieu-Jones, J.H. Moore, P.R.O.-A.A.L.S.C.T. CONSORTIUM, Missing data imputation in the electronic health record using deeply learned autoencoders, in: Pacific Symposium on Biocomputing 2017, World Scientific, 2017: pp. 207–218.
[75]B. Xue, D. Li, C. Lu, C.R. King, T. Wildes, et al., Use of Machine Learning to Develop and Evaluate Models Using Preoperative and Intraoperative Data to Identify Risks of Postoperative Complications, JAMA Netw Open. 4 (2021). https://doi.org/10.1001/jamanetworkopen.2021.2240.
[76]M.M. Ali, B.K. Paul, K. Ahmed, F.M. Bui, J.M.W. Quinn, et al., Heart disease prediction using supervised machine learning algorithms: Performance analysis and comparison, Comput Biol Med. 136 (2021) 104672. https://doi.org/10.1016/j.compbiomed.2021.104672.
[77]N.A. Azit, S. Sahran, V.M. Leow, M. Subramaniam, S. Mokhtar, et al., Prediction of hepatocellular carcinoma risk in patients with type-2 diabetes using supervised machine learning classification model, Heliyon. 8 (2022) e10772. https://doi.org/10.1016/j.heliyon.2022.e10772.
[78]S. Uddin, A. Khan, M.E. Hossain, M.A. Moni, Comparing different supervised machine learning algorithms for disease prediction, BMC Med Inform Decis Mak. 19 (2019) 281. https://doi.org/10.1186/s12911-019-1004-8.
[79]C.M. Lynch, B. Abdollahi, J.D. Fuqua, A.R. de Carlo, J.A. Bartholomai, et al., Prediction of lung cancer patient survival via supervised machine learning classification techniques, Int J Med Inform. 108 (2017) 1–8. https://doi.org/10.1016/j.ijmedinf.2017.09.013.