{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/neural-networks-versus-logistic-regression","title":"Neural networks versus Logistic regression for 30 days all-cause readmission prediction","arxiv_id":"1812.09549","date":"2018-12-22","proceeding":null,"authors":["Ahmed Allam","Mate Nagy","George Thoma","Michael Krauthammer"],"abstract":"Heart failure (HF) is one of the leading causes of hospital admissions in the\nUS. Readmission within 30 days after a HF hospitalization is both a recognized\nindicator for disease progression and a source of considerable financial burden\nto the healthcare system. Consequently, the identification of patients at risk\nfor readmission is a key step in improving disease management and patient\noutcome. In this work, we used a large administrative claims dataset to\n(1)explore the systematic application of neural network-based models versus\nlogistic regression for predicting 30 days all-cause readmission after\ndischarge from a HF admission, and (2)to examine the additive value of\npatients' hospitalization timelines on prediction performance. Based on data\nfrom 272,778 (49% female) patients with a mean (SD) age of 73 years (14) and\n343,328 HF admissions (67% of total admissions), we trained and tested our\npredictive readmission models following a stratified 5-fold cross-validation\nscheme. Among the deep learning approaches, a recurrent neural network (RNN)\ncombined with conditional random fields (CRF) model (RNNCRF) achieved the best\nperformance in readmission prediction with 0.642 AUC (95% CI, 0.640-0.645).\nOther models, such as those based on RNN, convolutional neural networks and CRF\nalone had lower performance, with a non-timeline based model (MLP) performing\nworst. A competitive model based on logistic regression with LASSO achieved a\nperformance of 0.643 AUC (95%CI, 0.640-0.646). We conclude that data from\npatient timelines improve 30 day readmission prediction for neural\nnetwork-based models, that a logistic regression with LASSO has equal\nperformance to the best neural network model and that the use of administrative\ndata result in competitive performance compared to published approaches based\non richer clinical datasets.","url_abs":"http://arxiv.org/abs/1812.09549v1","url_pdf":"http://arxiv.org/pdf/1812.09549v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"neural-networks-versus-logistic-regression","repo_url":"https://bitbucket.org/A_2/hcup_research","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"all","task_name":"All"},{"task_slug":"management","task_name":"Management"},{"task_slug":"readmission-prediction","task_name":"Readmission Prediction"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[{"method_slug":"logistic-regression","method_name":"Logistic Regression"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}