{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,9]],"date-time":"2026-05-09T08:10:43Z","timestamp":1778314243923,"version":"3.51.4"},"reference-count":35,"publisher":"SAGE Publications","license":[{"start":{"date-parts":[[2026,5,6]],"date-time":"2026-05-06T00:00:00Z","timestamp":1778025600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/journals.sagepub.com\/page\/policies\/text-and-data-mining-license"}],"content-domain":{"domain":["journals.sagepub.com"],"crossmark-restriction":true},"short-container-title":["J Intensive Care Med"],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Accurate prediction of short-term mortality in sepsis patients is critical for timely clinical decision-making. However, existing deep learning models often focus on static physiological parameters while neglecting the dynamic response to medical interventions, leading to risk underestimation due to the \u201cmasking effect\u201d of therapeutic measures.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We propose a lightweight hybrid deep learning framework that integrates dynamic intervention responses to predict 24-h all-cause mortality. Utilizing the MIMIC-IV v3.1 database, we included 13,788 adult sepsis patients. The model employs a dual-branch architecture: a Bidirectional LSTM to capture local temporal trends and a Transformer Encoder to extract global long-range dependencies. Crucially, we constructed a high-resolution feature set that includes vasopressor infusion rates and hourly urine output to quantify physiological feedback to resuscitation.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The proposed model achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.8139, significantly outperforming seven mainstream baselines, including LightGBM (0.8015), Bi-LSTM (0.7870), and pure Transformer models (0.7704). Feature importance analysis revealed that indicators of treatment response, specifically urine output and norepinephrine dosage, were among the top predictive features, validating the clinical hypothesis that drug dependency and renal perfusion are sensitive markers of prognosis. Furthermore, external validation on the independent multi-center eICU Collaborative Research Database demonstrated robust generalizability: a zero-shot transfer yielded an AUROC of 0.6620, which improved to 0.7347 after lightweight domain adaptation fine-tuning, with a Negative Predictive Value (NPV) of 90.04%, confirming the model's cross-institutional applicability as a reliable rule-out tool.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>Our LSTM-Transformer Fusion architecture effectively captures the complex \u201cdrug-physiology\u201d interactions with low computational cost. By explicitly modeling the dynamic response to treatment and demonstrating cross-institutional generalizability through external validation on the eICU database, this lightweight model offers a robust and interpretable tool for early warning systems in resource-constrained intensive care environments.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1177\/08850666261447860","type":"journal-article","created":{"date-parts":[[2026,5,6]],"date-time":"2026-05-06T13:29:36Z","timestamp":1778074176000},"update-policy":"https:\/\/doi.org\/10.1177\/sage-journals-update-policy","source":"Crossref","is-referenced-by-count":0,"title":["A Lightweight LSTM-Transformer Fusion Architecture for Real-Time Sepsis Mortality Prediction"],"prefix":"10.1177","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-4754-6534","authenticated-orcid":false,"given":"Zekai","family":"Yu","sequence":"first","affiliation":[{"name":"Hangzhou Dianzi University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Feiwei","family":"Qin","sequence":"additional","affiliation":[{"name":"Hangzhou Dianzi University"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhu","family":"Zhu","sequence":"additional","affiliation":[{"name":"Zhejiang University School of Medicine"},{"name":"Sino-Finland Joint AI Laboratory for Child Health of Zhejiang Province, Hangzhou, Zhejiang Province 310052, PR China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"179","published-online":{"date-parts":[[2026,5,6]]},"reference":[{"key":"e_1_3_3_2_2","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2016.0287"},{"key":"e_1_3_3_3_2","doi-asserted-by":"publisher","DOI":"10.1016\/S0140-6736(19)32989-7"},{"key":"e_1_3_3_4_2","doi-asserted-by":"publisher","DOI":"10.1007\/s00134-021-06506-y"},{"key":"e_1_3_3_5_2","doi-asserted-by":"publisher","DOI":"10.1007\/s00134-019-05872-y"},{"key":"e_1_3_3_6_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-018-0316-z"},{"key":"e_1_3_3_7_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41597-019-0103-9"},{"key":"e_1_3_3_8_2","doi-asserted-by":"publisher","DOI":"10.1109\/RBME.2020.3007816"},{"key":"e_1_3_3_9_2","doi-asserted-by":"publisher","DOI":"10.3389\/fmed.2021.607952"},{"issue":"1","key":"e_1_3_3_10_2","article-title":"Attend and diagnose: Clinical time series analysis using attention models","volume":"32","author":"Song H","unstructured":"Song H, Rajan D, Thiagarajan J, Spanias A. Attend and diagnose: Clinical time series analysis using attention models. Proceedings of the AAAI conference on artificial intelligence. 2018, April;32(1).","journal-title":"Proceedings of the AAAI conference on artificial intelligence"},{"key":"e_1_3_3_11_2","doi-asserted-by":"publisher","DOI":"10.1097\/CCM.0000000000004476"},{"key":"e_1_3_3_12_2","doi-asserted-by":"publisher","DOI":"10.1186\/cc10318"},{"key":"e_1_3_3_13_2","unstructured":"Johnson A Bulgarelli L Pollard T et al. 2024. MIMIC-IV (version 3.1). PhysioNet. RRID:SCR_007345. https:\/\/doi.org\/10.13026\/kpb9-mt58."},{"key":"e_1_3_3_14_2","doi-asserted-by":"publisher","DOI":"10.1161\/01.cir.101.23.e215"},{"key":"e_1_3_3_15_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-018-24271-9"},{"issue":"1","key":"e_1_3_3_16_2","first-page":"1929","article-title":"Dropout: A simple way to prevent neural networks from overfitting","volume":"15","author":"Srivastava N","year":"2014","unstructured":"Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout: A simple way to prevent neural networks from overfitting. The Journal of Machine Learning Research. 2014;15(1):1929-1958.","journal-title":"The Journal of Machine Learning Research"},{"key":"e_1_3_3_17_2","article-title":"Pytorch: An imperative style, high-performance deep learning library","volume":"32","author":"Paszke A","year":"2019","unstructured":"Paszke A, Gross S, Massa F, et al. Pytorch: An imperative style, high-performance deep learning library. Adv Neural Inf Process Syst. 2019;32.","journal-title":"Adv Neural Inf Process Syst."},{"key":"e_1_3_3_18_2","unstructured":"Loshchilov I Hutter F. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101. https:\/\/doi.org\/10.48550\/arXiv.1711.05101."},{"key":"e_1_3_3_19_2","first-page":"2980","article-title":"Focal loss for dense object detection","author":"Lin TY","year":"2017","unstructured":"Lin TY, Goyal P, Girshick R, He K, Doll\u00e1r P. Focal loss for dense object detection. Proceedings of the IEEE International Conference on Computer Vision. 2017:2980-2988.","journal-title":"Proceedings of the IEEE International Conference on Computer Vision"},{"key":"e_1_3_3_20_2","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0118432"},{"key":"e_1_3_3_21_2","doi-asserted-by":"publisher","DOI":"10.1002\/1097-0142(1950)3:1<32::aid-cncr2820030106>3.0.co;2-3"},{"key":"e_1_3_3_22_2","article-title":"Lightgbm: A highly efficient gradient boosting decision tree","volume":"30","author":"Ke G","year":"2017","unstructured":"Ke G, Meng Q, Finley T, et al. Lightgbm: A highly efficient gradient boosting decision tree. Adv Neural Inf Process Syst. 2017;30.","journal-title":"Adv Neural Inf Process Syst."},{"key":"e_1_3_3_23_2","doi-asserted-by":"publisher","DOI":"10.1145\/2939672.2939785"},{"key":"e_1_3_3_24_2","doi-asserted-by":"publisher","DOI":"10.1023\/A:1010933404324"},{"key":"e_1_3_3_25_2","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1997.9.8.1735"},{"key":"e_1_3_3_26_2","doi-asserted-by":"crossref","unstructured":"Cho K Van Merri\u00ebnboer B Gulcehre C et al. 2014. Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078. https:\/\/doi.org\/10.48550\/arXiv.1406.1078.","DOI":"10.3115\/v1\/D14-1179"},{"key":"e_1_3_3_27_2","article-title":"Attention is all you need","volume":"30","author":"Vaswani A","year":"2017","unstructured":"Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need. Adv Neural Inf Process Syst. 2017;30.","journal-title":"Adv Neural Inf Process Syst."},{"key":"e_1_3_3_28_2","article-title":"A unified approach to interpreting model predictions","volume":"30","author":"Lundberg SM","year":"2017","unstructured":"Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Adv Neural Inf Process Syst. 2017;30.","journal-title":"Adv Neural Inf Process Syst."},{"key":"e_1_3_3_29_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-018-0213-5"},{"key":"e_1_3_3_30_2","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMra1208943"},{"key":"e_1_3_3_31_2","doi-asserted-by":"publisher","DOI":"10.1186\/s12874-023-02138-6"},{"key":"e_1_3_3_32_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-021-00455-y"},{"key":"e_1_3_3_33_2","unstructured":"Pollard T Johnson A Raffa J Celi LA Badawi O Mark R. 2019. eICU Collaborative Research Database (version 2.0). PhysioNet. RRID:SCR_007345. https:\/\/doi.org\/10.13026\/C2WM1R."},{"key":"e_1_3_3_34_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijcard.2025.133993"},{"key":"e_1_3_3_35_2","doi-asserted-by":"publisher","DOI":"10.1111\/dom.70346"},{"key":"e_1_3_3_36_2","doi-asserted-by":"publisher","DOI":"10.1007\/s13167-025-00422-7"}],"container-title":["Journal of Intensive Care Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/journals.sagepub.com\/doi\/pdf\/10.1177\/08850666261447860","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/journals.sagepub.com\/doi\/full-xml\/10.1177\/08850666261447860","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/journals.sagepub.com\/doi\/pdf\/10.1177\/08850666261447860","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,9]],"date-time":"2026-05-09T07:21:10Z","timestamp":1778311270000},"score":1,"resource":{"primary":{"URL":"https:\/\/journals.sagepub.com\/doi\/10.1177\/08850666261447860"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,6]]},"references-count":35,"alternative-id":["10.1177\/08850666261447860"],"URL":"https:\/\/doi.org\/10.1177\/08850666261447860","relation":{},"ISSN":["0885-0666","1525-1489"],"issn-type":[{"value":"0885-0666","type":"print"},{"value":"1525-1489","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,6]]},"article-number":"08850666261447860"}}