{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,5]],"date-time":"2026-11-05T10:15:51Z","timestamp":1793873751548,"version":"4.6.1"},"reference-count":21,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2021,1,8]],"date-time":"2021-01-08T00:00:00Z","timestamp":1610064000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2021,1,8]],"date-time":"2021-01-08T00:00:00Z","timestamp":1610064000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"DOI":"10.13039\/100000050","name":"U.S. Department of Health & Human Services | NIH | National Heart, Lung, and Blood Institute","doi-asserted-by":"publisher","award":["R01 HL144692"],"award-info":[{"award-number":["R01 HL144692"]}],"id":[{"id":"10.13039\/100000050","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["npj Digit. Med."],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>\n                    While deep neural networks (DNNs) and other machine learning models often have higher accuracy than simpler models like logistic regression (LR), they are often considered to be \u201cblack box\u201d models and this lack of interpretability and transparency is considered a challenge for clinical adoption. In healthcare, intelligible models not only help clinicians to understand the problem and create more targeted action plans, but also help to gain the clinicians\u2019 trust. One method of overcoming the limited interpretability of more complex models is to use Generalized Additive Models (GAMs). Standard GAMs simply model the target response as a sum of univariate models. Inspired by GAMs, the same idea can be applied to neural networks through an architecture referred to as Generalized Additive Models with Neural Networks (GAM-NNs). In this manuscript, we present the development and validation of a model applying the concept of GAM-NNs to allow for interpretability by visualizing the learned feature patterns related to risk of in-hospital mortality for patients undergoing surgery under general anesthesia. The data consists of 59,985 patients with a feature set of 46 features extracted at the end of surgery to which we added previously not included features: total anesthesia case time (1 feature); the time in minutes spent with mean arterial pressure (MAP) below 40, 45, 50, 55, 60, and 65\u2009mmHg during surgery (6 features); and Healthcare Cost and Utilization Project (HCUP) Code Descriptions of the Primary current procedure terminology (CPT) codes (33 features) for a total of 86 features. All data were randomly split into 80% for training (\n                    <jats:italic>n<\/jats:italic>\n                    \u2009=\u200947,988) and 20% for testing (\n                    <jats:italic>n<\/jats:italic>\n                    \u2009=\u200911,997) prior to model development. Model performance was compared to a standard LR model using the same features as the GAM-NN. The data consisted of 59,985 surgical records, and the occurrence of in-hospital mortality was 0.81% in the training set and 0.72% in the testing set. The GAM-NN model with HCUP features had the highest area under the curve (AUC) 0.921 (0.895\u20130.95). Overall, both GAM-NN models had higher AUCs than LR models, however, had lower average precisions. The LR model without HCUP features had the highest average precision 0.217 (0.136\u20130.31). To assess the interpretability of the GAM-NNs, we then visualized the learned contributions of the GAM-NNs and compared against the learned contributions of the LRs for the models with HCUP features. Overall, we were able to demonstrate that our proposed generalized additive neural network (GAM-NN) architecture is able to (1) leverage a neural network\u2019s ability to learn nonlinear patterns in the data, which is more clinically intuitive, (2) be interpreted easily, making it more clinically useful, and (3) maintain model performance as compared to previously published DNNs.\n                  <\/jats:p>","DOI":"10.1038\/s41746-020-00377-1","type":"journal-article","created":{"date-parts":[[2021,1,8]],"date-time":"2021-01-08T06:19:34Z","timestamp":1610086774000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":59,"title":["Development and validation of an interpretable neural network for prediction of postoperative in-hospital mortality"],"prefix":"10.1038","volume":"4","author":[{"given":"Christine K.","family":"Lee","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Muntaha","family":"Samad","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ira","family":"Hofer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2120-1672","authenticated-orcid":false,"given":"Maxime","family":"Cannesson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pierre","family":"Baldi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2021,1,8]]},"reference":[{"key":"377_CR1","doi-asserted-by":"publisher","first-page":"877","DOI":"10.1016\/j.bja.2019.07.030","volume":"123","author":"BL Hill","year":"2019","unstructured":"Hill, B. L. et al. An automated machine learning-based model predicts postoperative mortality using readily-extractable preoperative electronic health record data. Br. J. Anaesth. 123, 877\u2013886 (2019).","journal-title":"Br. J. Anaesth."},{"key":"377_CR2","doi-asserted-by":"publisher","first-page":"688","DOI":"10.1016\/j.bja.2019.07.025","volume":"123","author":"BA Fritz","year":"2019","unstructured":"Fritz, B. A. et al. Deep-learning model for predicting 30-day postoperative mortality. Br. J. Anaesth. 123, 688\u2013695 (2019).","journal-title":"Br. J. Anaesth."},{"key":"377_CR3","doi-asserted-by":"publisher","first-page":"649","DOI":"10.1097\/ALN.0000000000002186","volume":"129","author":"CK Lee","year":"2018","unstructured":"Lee, C. K., Hofer, I., Gabel, E., Baldi, P. & Cannesson, M. Development and validation of a deep neural network model for prediction of postoperative in-hospital mortality. Anesthesiology 129, 649\u2013662 (2018).","journal-title":"Anesthesiology"},{"key":"377_CR4","doi-asserted-by":"publisher","first-page":"1419","DOI":"10.1093\/jamia\/ocy068","volume":"25","author":"C Xiao","year":"2018","unstructured":"Xiao, C., Choi, E. & Sun, J. Opportunities and challenges in developing deep learning models using electronic health records data: a systematic review. J. Am. Med. Inform. Assoc. 25, 1419\u20131428 (2018).","journal-title":"J. Am. Med. Inform. Assoc."},{"key":"377_CR5","doi-asserted-by":"publisher","first-page":"224","DOI":"10.1186\/s13054-015-0932-7","volume":"19","author":"JL Vincent","year":"2015","unstructured":"Vincent, J. L. et al. Perioperative cardiovascular monitoring of high-risk patients: a consensus of 12. Crit. Care 19, 224 (2015).","journal-title":"Crit. Care"},{"key":"377_CR6","doi-asserted-by":"publisher","unstructured":"Caruana, R. et al. Intelligible models for HealthCare. In Proc. 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD \u201915 1721\u20131730 (ACM Press, 2015), https:\/\/doi.org\/10.1145\/2783258.2788613.","DOI":"10.1145\/2783258.2788613"},{"key":"377_CR7","doi-asserted-by":"publisher","unstructured":"Potts, W. J. E. Generalized additive neural networks. In Proc. Fifth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 194\u2013200 (Association for Computing Machinery (ACM), NY, 1999), https:\/\/doi.org\/10.1145\/312129.312228.","DOI":"10.1145\/312129.312228"},{"key":"377_CR8","doi-asserted-by":"publisher","first-page":"719","DOI":"10.1007\/s00521-017-3105-6","volume":"31","author":"C Br\u00e1s-Geraldes","year":"2019","unstructured":"Br\u00e1s-Geraldes, C., Papoila, A. & Xufre, P. Generalized additive neural network with flexible parametric link function: model estimation using simulated and real clinical data. Neural Comput. Appl. 31, 719\u2013736 (2019).","journal-title":"Neural Comput. Appl."},{"key":"377_CR9","doi-asserted-by":"publisher","first-page":"181","DOI":"10.1146\/annurev-biodatasci-080917-013343","volume":"1","author":"P Baldi","year":"2018","unstructured":"Baldi, P. Deep learning in biomedical data science. Annu. Rev. Biomed. Data Sci. 1, 181\u2013205 (2018).","journal-title":"Annu. Rev. Biomed. Data Sci."},{"key":"377_CR10","doi-asserted-by":"publisher","first-page":"1069","DOI":"10.1053\/j.gastro.2018.06.037","volume":"155","author":"G Urban","year":"2018","unstructured":"Urban, G. et al. Deep learning localizes and identifies polyps in real time with 96% accuracy in screening colonoscopy. Gastroenterology 155, 1069\u20131078.e8 (2018).","journal-title":"Gastroenterology"},{"key":"377_CR11","doi-asserted-by":"publisher","first-page":"1201","DOI":"10.3174\/ajnr.A5667","volume":"39","author":"P Chang","year":"2018","unstructured":"Chang, P. et al. Deep-learning convolutional neural networks accurately classify genetic mutations in gliomas. Am. J. Neuroradiol. 39, 1201\u20131207 (2018).","journal-title":"Am. J. Neuroradiol."},{"key":"377_CR12","doi-asserted-by":"publisher","first-page":"e15154","DOI":"10.2196\/15154","volume":"22","author":"O Asan","year":"2020","unstructured":"Asan, O., Bayrak, A.E. & Choudhury, A. Artificial intelligence and human trust in healthcare: focus on clinicians. J. Med. Internet Res. 22, e15154 (2020).","journal-title":"J. Med. Internet Res."},{"key":"377_CR13","first-page":"359","volume":"106","author":"S Tonekaboni","year":"2019","unstructured":"Tonekaboni, S., Joshi, S., McCradden, M.D. & Goldenberg, A. What clinicians want: contextualizing explainable machine learning for clinical end use. Proc. Mach. Learn. Res. 106, 359\u2013380 (2019).","journal-title":"Proc. Mach. Learn. Res."},{"key":"377_CR14","doi-asserted-by":"publisher","first-page":"1477","DOI":"10.1097\/CCM.0000000000003803","volume":"47","author":"JC Ginestra","year":"2019","unstructured":"Ginestra, J. C. et al. Clinician perception of a machine learning-based early warning system designed to predict severe sepsis and septic shock. Crit. Care Med. 47, 1477\u20131484 (2019).","journal-title":"Crit. Care Med."},{"key":"377_CR15","doi-asserted-by":"publisher","first-page":"1880","DOI":"10.1213\/ANE.0000000000001201","volume":"122","author":"IS Hofer","year":"2016","unstructured":"Hofer, I. S., Gabel, E., Pfeffer, M., Mahbouba, M. & Mahajan, A. A systematic approach to creation of a perioperative data warehouse. Anesth. Analg. 122, 1880\u20131884 (2016).","journal-title":"Anesth. Analg."},{"key":"377_CR16","doi-asserted-by":"publisher","first-page":"e323","DOI":"10.2196\/jmir.5870","volume":"18","author":"W Luo","year":"2016","unstructured":"Luo, W. et al. Guidelines for developing and reporting machine learning predictive models in biomedical research: a multidisciplinary view. J. Med. Internet Res. 18, e323 (2016).","journal-title":"J. Med. Internet Res."},{"key":"377_CR17","unstructured":"Kingma, D. P. & Lei Ba, J. Adam: a method for stochastic optimization. Preprint at https:\/\/arxiv.org\/abs\/1412.6980."},{"key":"377_CR18","first-page":"1929","volume":"15","author":"N Srivastava","year":"2014","unstructured":"Srivastava, N., Hinton, G., Krizhevsky, A. & Salakhutdinov, R. Dropout: a simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 15, 1929\u20131958 (2014).","journal-title":"J. Mach. Learn. Res."},{"key":"377_CR19","doi-asserted-by":"publisher","first-page":"78","DOI":"10.1016\/j.artint.2014.02.004","volume":"210","author":"P Baldi","year":"2014","unstructured":"Baldi, P. & Sadowski, P. The dropout learning algorithm. Artif. Intell. 210, 78\u2013122 (2014).","journal-title":"Artif. Intell."},{"key":"377_CR20","unstructured":"Cholett, F. Keras. https:\/\/keras.io (2015). Accessed 19 Dec 2020."},{"key":"377_CR21","first-page":"2825","volume":"12","author":"F Pedregosa","year":"2011","unstructured":"Pedregosa, F. et al. Scikit-learn: machine learning in Python. J. Mach. Learn. Res. 12, 2825\u20132830 (2011).","journal-title":"J. Mach. Learn. Res."}],"container-title":["npj Digital Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41746-020-00377-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41746-020-00377-1","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41746-020-00377-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,12,3]],"date-time":"2022-12-03T14:09:47Z","timestamp":1670076587000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41746-020-00377-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,1,8]]},"references-count":21,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2021,12]]}},"alternative-id":["377"],"URL":"https:\/\/doi.org\/10.1038\/s41746-020-00377-1","relation":{},"ISSN":["2398-6352"],"issn-type":[{"value":"2398-6352","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,1,8]]},"assertion":[{"value":"17 August 2020","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"25 November 2020","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"8 January 2021","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"Christine Lee is currently a full-time employee at Edwards Lifesciences, however, this work was done independent from her employer and as a part of her PhD. Maxime Cannesson: Ownership interest in Sironis, a company developing closed-loop systems; Consulting for Edwards Lifesciences (Irvine, CA) and Masimo Corp. (Irvine, CA). Maxime Cannesson has received research support from Edwards Lifesciences through his Department. The other authors declare no competing interests concerning this article.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Competing interests"}}],"article-number":"8"}}