Evaluating a Recurrent Neural Network Model for Predicting Readmission to Cardiovascular ICUs Based on Clinical Time Series Data †
Abstract
:1. Introduction
2. Materials and Methods
- Patients who were transferred from the ICU to a normal station and then returned to the ICU within 48 h. Due to possible logistical reasons, patients whose first ICU stay was less than 24 h are excluded. These patients are labeled as ’readmitted’.
- Patients who died during the hospital stay. The dead subjects are also labeled as ’readmitted’.
- Patients who were transferred from the ICU to a normal station and then were not returned to the ICU within 30 days. Those who did not return within 30 days after transfer from ICU are labeled as ’non-readmitted’.
2.1. Patient Cohorts
2.2. Time Series and Patient Data
2.3. Preprocessing
2.4. Model Training
2.5. Model Validation
2.6. Model Evaluation
3. Results
3.1. Model Training and Validation
3.2. Model Evaluation
4. Discussion
5. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
Abbreviations
UKD | University Hospital Düsseldorf |
ICU | Intensive Care Unit |
RNN | Recurrent Neural Network |
LSTM | Long Short-Term Memory |
ML | Machine Learning |
AI | Artificial Intelligence |
MIMIC | Medical Information Mart for Intensive Care |
BRITS | Bidirectional Recurrent Imputation for Time Series |
XAI | eXplainable AI |
References
- Rosenberg, A.L.; Charles, W. Patients Readmitted to ICUs: A Systematic Review of Risk Factors and Outcomes. Chest 2000, 118, 492–502. [Google Scholar] [CrossRef] [PubMed]
- van Sluisveld, N.; Bakhshi-Raiez, F.; de Keizer, N.; Holman, R.; Wester, G.; Wollersheim, H.; van der Hoven, J.G.; Zegers, M. Variation in rates of ICU readmissions and post-ICU in-hospital mortality and their association with ICU discharge practices. BMC Health Serv. Res. 2017, 17, 281. [Google Scholar] [CrossRef] [PubMed]
- Desautels, T.; Das, R.; Calvert, J.; Trivedi, M.; Summers, C.; Wales, D.J.; Ercole, A. Prediction of early unplanned intensive care unit readmission in a UK tertiary care hospital: A cross-sectional machine learning approach. BMJ Open 2017, 7, e017199. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Lin, W.T.; Chen, W.L.; Chao, C.M.; Lai, C.C. The outcomes and prognostic factors of the patients with unplanned intensive care unit readmissions. Medicine 2018, 97, e11124. [Google Scholar] [CrossRef] [PubMed]
- Azarfarin, R.; Ashouri, N.; Totonchi, Z.; Bakhshandeh, H.; Yaghoubi, A. Factors influencing prolonged ICU stay after open heart surgery. Res. Cardiovasc. Med. 2014, 3, e20159. [Google Scholar] [CrossRef] [PubMed]
- Almashrafi, A.; Elmontsri, M.; Aylin, P. Systematic review of factors influencing length of stay in ICU after adult cardiac surgery. BMC Health Serv. Res. 2016, 16, 318. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Andrearczyk, V.; Fontaine, P.; Oreiller, V.; Castelli, J.; Jreige, M.; Prior., J.O.; Depeursinge, A. Multi-task Deep Segmentation and Radiomics for Automatic Prognosis in Head and Neck Cancer. In Predictive Intelligence in Medicine. PRIME 2021. Lecture Notes in Computer Science; Rekik, I., Adeli, E., Park, S.H., Schnabel, J., Eds.; Springer: Cham, Switzerland, 2021. [Google Scholar] [CrossRef]
- Moazemi, S.; Essler, M.; Schultz, T.; Bundschuh, R.A. Predicting Treatment Response in Prostate Cancer Patients Based on Multimodal PET/CT for Clinical Decision Support. In Multimodal Learning for Clinical Decision Support. ML-CDS 2021. Lecture Notes in Computer Science; Springer: Cham, Switzerland, 2021. [Google Scholar] [CrossRef]
- Khatami, M.; Wehler, R.; Schultz, T. Parcellation-Free prediction of task fMRI activations from dMRI tractography. Med. Image Anal. 2022, 76, 102317. [Google Scholar] [CrossRef] [PubMed]
- Khan, A.R.; Wang, L.; Beg, M.F. FreeSurfer-Initiated Fully-Automated Subcortical Brain Segmentation in MRI Using Large Deformation Diffeomorphic Metric Mapping. NeuroImage 2008, 41, 735–746. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Hsu, G.C. Using Time-Series and Forecasting to Manage Type 2 Diabetes Conditions (GH-Method: Math-Physical Medicine); Universdad de Granada: Granada, Spain, 2019. [Google Scholar]
- Spiga, R.; Batton-Hubert, M.; Sarazin, M. Predicting hospital admissions with integer-valued time series. In ITISE 2019. Proceedings of Papers; ITISE: Grenade, Spain, 2019; pp. 897–998. [Google Scholar]
- Šimpach, O.; Dotlačilová, P. Age-Specific Death Rates Smoothed by the Gompertz–Makeham Function and Their Application in Projections by Lee–Carter Model. In Time Series Analysis and Forecasting. Contributions to Statistics; Rojas, I., Pomares, H., Eds.; Springer: Cham, Switzerland, 2016. [Google Scholar] [CrossRef]
- Lin, Y.W.; Zhou, Y.; Faghri, F.; Shaw, M.J.; Campbell, R.H. Analysis and prediction of unplanned intensive care unit readmission using recurrent neural networks with long short-term memory. PLoS ONE 2019, 14, e0218942. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Huang, Y.; Talwar, A.; Chatterjee, S.; Rajender, R.A. Application of machine learning in predicting hospital readmissions: A scoping review of the literature. BMC Med. Res. Methodol. 2021, 21, 96. [Google Scholar] [CrossRef] [PubMed]
- Ashfaq, A.; Sant’Anna, A.; Lingman, M.; Nowaczyk, S. Readmission prediction using deep learning on electronic health records. J. Biomed. Inform. 2019, 97, 103256. [Google Scholar] [CrossRef] [PubMed]
- Kessler, S.; Schroeder, D.; Korlakov, S.; Hettlich, V.; Lichtenberg, A.; Schmid, F.; Aubin, H. Predicting readmission to the cardiovascular ICU using recurrent neural networks. J. Digital. Health 2022. manucript in final decision. [Google Scholar]
- Johnson, A.; Pollard, T.; Shen, L.; Lehman, L.H.; Feng, M.; Ghassemi, M.; Moody, B.; Szolovits, P.; Celi, L.A.; Mark, R.G. MIMIC-III, a freely accessible critical care database. Sci. Data 2016, 3, 160035. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Paszke, A.; Gross, S.; Massa, F.; Lerer, A.; Bradbury, J.; Chanan, G.; Killeen, T.; Lin, Z.; Gimelsshein, N.; Antiga, L.; et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library. In Proceedings of the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada, 8–14 December 2019; pp. 8024–8035. [Google Scholar]
- Bauder, R.A.; Khoshgoftaar, T.M.; Hasanin, T. An Empirical Study on Class Rarity in Big Data. In Proceedings of the 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA), Orlando, FL, USA, 17–20 December 2018; pp. 785–790. [Google Scholar] [CrossRef]
- Brown, S.E.; Ratcliffe, S.J.; Halpern, S.D. An empirical derivation of the optimal time interval for defining ICU readmissions. Med. Care 2013, 51, 706–714. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Gudivada, V.; Apon, A.; Ding, J. Data Quality Considerations for Big Data and Machine Learning: Going Beyond Data Cleaning and Transformations. Int. J. Adv. Softw. 2017, 10, 1–20. [Google Scholar]
- Cao, W.; Wang, D.; Li, J.; Zhou, H.; Li, L.; Li, Y. BRITS: Bidirectional Recurrent Imputation for Time Series. Adv. Neural Inf. Process. Syst. 2018, 31, 6775–6785. [Google Scholar]
- Quiñonero-Candela, J.; Sugiyama, M.; Schwaighofer, A.; Lawrence, N.D.; Jordan, M.I.; Dietterich, T.G. Dataset Shift in Machine Learning; MIT Press: Cambridge, MA, USA, 2008; ISBN 978-0-262-17005-5. [Google Scholar]
Lab Values | Vital Signs | Patient Information |
---|---|---|
Creatinine | Body Temperature | Age |
Blood PH | ABP | Weight |
Sodium | Heart Rate | LoS |
Potassium | Oxygenation | |
Hematocrit | ||
Bilirubin |
Hyperparameters | Values |
---|---|
optimization algorithm | Adam |
learning rate | 0.003 |
loss function | cross-entropy |
batch size | 32 |
epochs | early stopping |
Balanced Accuracy | Recall | Precision | AUC-PR | AUC-ROC | |
---|---|---|---|---|---|
cropped to 48 h | 0.799 | 0.694 | 0.389 | 0.666 | 0.877 |
uncropped | 0.778 | 0.649 | 0.389 | 0.636 | 0.859 |
UKD | |||||
---|---|---|---|---|---|
Balanced Accuracy | Recall | Precision | AUC-PR | AUC-ROC | |
cropped to 48 h | 0.514 | 0.06 | 0.318 | 0.223 | 0.554 |
uncropped | 0.498 | 0.22 | 0.196 | 0.221 | 0.517 |
MIMIC | |||||
Balanced Accuracy | Recall | Precision | AUC-PR | AUC-ROC | |
cropped to 48 h | 0.74 | 0.575 | 0.35 | 0.538 | 0.796 |
uncropped | 0.766 | 0.628 | 0.477 | 0.571 | 0.828 |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Moazemi, S.; Kalkhoff, S.; Kessler, S.; Boztoprak, Z.; Hettlich, V.; Liebrecht, A.; Bibo, R.; Dewitz, B.; Lichtenberg, A.; Aubin, H.; et al. Evaluating a Recurrent Neural Network Model for Predicting Readmission to Cardiovascular ICUs Based on Clinical Time Series Data. Eng. Proc. 2022, 18, 1. https://doi.org/10.3390/engproc2022018001
Moazemi S, Kalkhoff S, Kessler S, Boztoprak Z, Hettlich V, Liebrecht A, Bibo R, Dewitz B, Lichtenberg A, Aubin H, et al. Evaluating a Recurrent Neural Network Model for Predicting Readmission to Cardiovascular ICUs Based on Clinical Time Series Data. Engineering Proceedings. 2022; 18(1):1. https://doi.org/10.3390/engproc2022018001
Chicago/Turabian StyleMoazemi, Sobhan, Sebastian Kalkhoff, Steven Kessler, Zeynep Boztoprak, Vincent Hettlich, Artur Liebrecht, Roman Bibo, Bastian Dewitz, Artur Lichtenberg, Hug Aubin, and et al. 2022. "Evaluating a Recurrent Neural Network Model for Predicting Readmission to Cardiovascular ICUs Based on Clinical Time Series Data" Engineering Proceedings 18, no. 1: 1. https://doi.org/10.3390/engproc2022018001