Benchmarking Deep Learning Architectures for Predicting Readmission to the ICU and Describing Patients-at-Risk

Autor: Sradha Kotwal, Angus Ritchie, Sebastiano Barbieri, James Kemp, Oscar Perez-Concha, Martin Gallagher, Louisa Jorm
Rok vydání: 2020
Předmět:
FOS: Computer and information sciences
0301 basic medicine
Computer Science - Machine Learning
Computer science
lcsh:Medicine
computer.software_genre
Medical care
Machine Learning (cs.LG)
law.invention
0302 clinical medicine
Statistics - Machine Learning
law
Odds Ratio
030212 general & internal medicine
Medical diagnosis
lcsh:Science
Interpretability
Multidisciplinary
Artificial neural network
Intensive care unit
Health services
Intensive Care Units
Disease Progression
Algorithms
Risk
Medical Records Systems
Computerized
Vital signs
Machine Learning (stat.ML)
Bayesian inference
Machine learning
Communicable Diseases
Patient Readmission
Sensitivity and Specificity
Article
03 medical and health sciences
Deep Learning
Humans
business.industry
Deep learning
lcsh:R
Bayes Theorem
Odds ratio
030104 developmental biology
Recurrent neural network
Chronic Disease
lcsh:Q
Neural Networks
Computer
Artificial intelligence
business
computer
Forecasting
Zdroj: Scientific Reports, Vol 10, Iss 1, Pp 1-10 (2020)
Scientific Reports
ISSN: 2045-2322
DOI: 10.1038/s41598-020-58053-z
Popis: To compare different deep learning architectures for predicting the risk of readmission within 30 days of discharge from the intensive care unit (ICU). The interpretability of attention-based models is leveraged to describe patients-at-risk. Several deep learning architectures making use of attention mechanisms, recurrent layers, neural ordinary differential equations (ODEs), and medical concept embeddings with time-aware attention were trained using publicly available electronic medical record data (MIMIC-III) associated with 45,298 ICU stays for 33,150 patients. Bayesian inference was used to compute the posterior over weights of an attention-based model. Odds ratios associated with an increased risk of readmission were computed for static variables. Diagnoses, procedures, medications, and vital signs were ranked according to the associated risk of readmission. A recurrent neural network, with time dynamics of code embeddings computed by neural ODEs, achieved the highest average precision of 0.331 (AUROC: 0.739, F1-Score: 0.372). Predictive accuracy was comparable across neural network architectures. Groups of patients at risk included those suffering from infectious complications, with chronic or progressive conditions, and for whom standard medical care was not suitable. Attention-based networks may be preferable to recurrent networks if an interpretable model is required, at only marginal cost in predictive accuracy.
Databáze: OpenAIRE
Externí odkaz: