ForecastICU: A Prognostic Decision Support System for Timely Prediction of Intensive Care Unit Admission

We develop ForecastICU: a prognostic decision support system that monitors hospitalized patients and prompts alarms for intensive care unit (ICU) admissions. ForecastICU is first trained in an offline stage by constructing a Bayesian belief system that corresponds to its belief about how trajectories of physiological data streams of the patient map to a clinical status. After that, ForecastICU monitors a new patient in real-time by observing her physiological data stream, updating its belief about her status over time, and prompting an alarm whenever its belief process hits a predefined threshold (confidence). Using a real-world dataset obtained from UCLA Ronald Reagan Medical Center, we show that ForecastICU can predict ICU admissions 9 hours before a physician's decision (for a sensitivity of 40% and a precision of 50%). Also, ForecastICU performs consistently better than other state-of-the-art machine learning algorithms in terms of sensitivity, precision, and timeliness: it can predict ICU admissions 3 hours earlier, and offers a 7.8% gain in sensitivity and a 5.1% gain in precision compared to the best state-of-the-art algorithm. Moreover, ForecastICU offers an area under curve (AUC) gain of 22.3% compared to the Rothman index, which is the currently deployed technology in most hospital wards.

[1]  Robert A Jacobs,et al.  Bayesian learning theory applied to human cognition. , 2011, Wiley interdisciplinary reviews. Cognitive science.

[2]  Lars Schmidt-Thieme,et al.  Time-Series Classification Based on Individualised Error Prediction , 2010, 2010 13th IEEE International Conference on Computational Science and Engineering.

[3]  Sylvie Chevret,et al.  Delayed intensive care unit admission is associated with increased mortality in patients with cancer with acute respiratory failure , 2013, Leukemia & lymphoma.

[4]  Li Wei,et al.  Semi-supervised time series classification , 2006, KDD '06.

[5]  Amaury Lendasse,et al.  Adaptive Ensemble Models of Extreme Learning Machines for Time Series Prediction , 2009, ICANN.

[6]  David A. Clifton,et al.  Modelling physiological deterioration in post-operative patient vital-sign data , 2013, Medical & Biological Engineering & Computing.

[7]  Peter Szolovits,et al.  A Multivariate Timeseries Modeling Approach to Severity of Illness Assessment and Forecasting in ICU with Sparse, Heterogeneous Clinical Data , 2015, AAAI.

[8]  Huan Liu,et al.  Feature Selection for High-Dimensional Data: A Fast Correlation-Based Filter Solution , 2003, ICML.

[9]  Peter Stagge,et al.  Recurrent neural networks for time series classification , 2003, Neurocomputing.

[10]  J. Bliss,et al.  The role of alarm signal duration as a cue for alarm validity. , 2007, Applied ergonomics.

[11]  Raouf N. Gorgui-Naguib,et al.  A fuzzy logic based-method for prognostic decision making in breast and prostate cancers , 2003, IEEE Transactions on Information Technology in Biomedicine.

[12]  J. Habbema,et al.  Prognostic modelling with logistic regression analysis: a comparison of selection and estimation methods in small data sets. , 2000, Statistics in medicine.

[13]  Vincent Liu,et al.  Adverse outcomes associated with delayed intensive care unit transfers in an integrated healthcare system. , 2012, Journal of hospital medicine.

[14]  Michael J. Rothman,et al.  Development and validation of a continuous measure of patient condition using the Electronic Medical Record , 2013, J. Biomed. Informatics.

[15]  Alok N. Choudhary,et al.  Twitter Trending Topic Classification , 2011, 2011 IEEE 11th International Conference on Data Mining Workshops.

[16]  E. Jansma,et al.  The impact of the use of the Early Warning Score (EWS) on patient outcomes: a systematic review. , 2014, Resuscitation.

[17]  A. Egol.  Guidelines for intensive care unit admission, discharge, and triage , 1999 .

[18]  G. Koop,et al.  Bayesian Multivariate Time Series Methods for Empirical Macroeconomics , 2009 .

[19]  C. Piech,et al.  Daily cost of an intensive care unit day: The contribution of mechanical ventilation* , 2005, Critical care medicine.

[20]  S. Bikhchandani,et al.  You have printed the following article : A Theory of Fads , Fashion , Custom , and Cultural Change as Informational Cascades , 2007 .

[21]  Mariea C. Dunn,et al.  Behavioural implications of alarm mistrust as a function of task workload , 2000, Ergonomics.

[22]  J. Tenenbaum,et al.  Theory-based Bayesian models of inductive learning and reasoning , 2006, Trends in Cognitive Sciences.

[23]  Elad Hazan,et al.  Online Time Series Prediction with Missing Data , 2015, ICML.

[24]  Philip S. Yu,et al.  Early classification on time series , 2012, Knowledge and Information Systems.

[25]  M. Fitzgerald,et al.  Review article: Shock Index for prediction of critical bleeding post‐trauma: A systematic review , 2014, Emergency medicine Australasia : EMA.

[26]  David A. Clifton,et al.  Multitask Gaussian Processes for Multivariate Physiological Time-Series Analysis , 2015, IEEE Transactions on Biomedical Engineering.

[27]  Mihaela van der Schaar,et al.  Forecasting Popularity of Videos Using Social Media , 2014, IEEE Journal of Selected Topics in Signal Processing.

[28]  Philip S. Yu,et al.  Early prediction on time series: a nearest neighbor approach , 2009, IJCAI 2009.

[29]  J. Habbema,et al.  Prognostic Modeling with Logistic Regression Analysis , 2001, Medical decision making : an international journal of the Society for Medical Decision Making.

[30]  Valerie J. Gooder,et al.  Inpatient transfers to the intensive care unit , 2003 .

[31]  Dev Jayaraman,et al.  Adverse outcomes associated with delayed intensive care consultation in medical and surgical inpatients. , 2012, Journal of critical care.

[32]  MS Dr. Michael P. Young MD,et al.  Inpatient transfers to the intensive care unit , 2007, Journal of General Internal Medicine.

[33]  O. Bouamra,et al.  The value of traditional vital signs, shock index, and age-based markers in predicting trauma mortality , 2013, The journal of trauma and acute care surgery.

[34]  Daoqiang Zhang,et al.  Multi-modal multi-task learning for joint prediction of multiple regression and classification variables in Alzheimer's disease , 2012, NeuroImage.

[35]  Jimeng Sun,et al.  Predicting Patient's Trajectory of Physiological Data using Temporal Trends in Similar Patients: A System for Near-Term Prognostics. , 2010, AMIA ... Annual Symposium proceedings. AMIA Symposium.

[36]  David A. Clifton,et al.  Gaussian process regression in vital-sign early warning systems , 2012, 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[37]  M. Cvach.  Monitor alarm fatigue: an integrative review. , 2012, Biomedical instrumentation & technology.

[38]  Peter Dayan,et al.  Bayesian Model of Behaviour in Economic Games , 2008, NIPS.

[39]  Xin Guo,et al.  Optimal stopping times with different information levels and with time uncertainty , 2007 .

[40]  Ben Glocker,et al.  Joint Classification-Regression Forests for Spatially Structured Multi-object Segmentation , 2012, ECCV.

[41]  Lei Li,et al.  Time Series Clustering: Complex is Simpler! , 2011, ICML.

[42]  C. Tsien,et al.  Poor prognosis for existing monitors in the intensive care unit. , 1997, Critical care medicine.

[43]  Jimeng Sun,et al.  A System for Mining Temporal Physiological Data Streams for Advanced Prognostic Decision Support , 2010, 2010 IEEE International Conference on Data Mining.

[44]  Arkaitz Zubiaga,et al.  Real‐time classification of Twitter trends , 2014, J. Assoc. Inf. Sci. Technol..

[45]  Gary B. Smith,et al.  ViEWS--Towards a national early warning score for detecting adult inpatient deterioration. , 2010, Resuscitation.

[46]  Simon Fong,et al.  Real-Time Clinical Decision Support System with Data Stream Mining , 2012, Journal of biomedicine & biotechnology.

[47]  Robert Gibbons,et al.  Using Electronic Health Record Data to Develop and Validate a Prediction Model for Adverse Outcomes in the Wards* , 2012, Critical care medicine.