Early Sepsis Prediction in Intensive Care Patients using Random Forest Classifier
DOI:
https://doi.org/10.26438/ijcse/v8i1.1722Keywords:
MachineLearning, EarlySepsisprediction, Random Forest Classifier, AUROCAbstract
Sepsis is one of the most common causes of morbidity and mortality in the Intensive Care Unit (ICU) patients. The lack of sensitive and specific clinical and laboratory variables for early identification of sepsis in critically ill patients is the causative factor for needless and delayed or untimely interruption of a proper antibiotic therapy. The current work developed a machine learning-based early sepsis prediction model in intensive care patients with vital parameters which evaluated the goodness of model fit and its accuracy. The predictors were extracted from combinations of vital sign measure and their changes over time. The dataset consisted of 20,336 patients (medical and surgical) who were admitted in ICU. Random Forest Classifier was used as the Machine learning algorithm for developing a predictive model. For the early prediction of sepsis in ICU patients, the Random Forest Classifier achieved an AUROC curve of 0.58 for the data collected from the patients within 24 hours. Sepsis is being the common cause of admission in ICU worldwide, a machine learning technique adopting statistical methods to conclude relationships between patient features and outcomes in large data set was successfully applied to predict adverse events.
References
[1] V. Ribas, A.Vellido, J. Rodriguez, J. Rello, “Severe sepsis mortality prediction with logistic regression over latent factors”, Expert Systems with Applications, pp. 1937-1943, 2012.
[2] S. Chatterjee, M. Bhattacharya, S. Todi, “Epidemiology of Adult population Sepsis in India: A single centre 5 year experience”, Indian Journal of Critical CareMedicine, pp. 35-39, 2017.
[3] A. Johnso, M. Ghassemi, S. Nemati, K. Niehaus, D. Clifton, G. Clifford, “Machine learning and Decision Support in Critical Care”, In the Proceedings of 2016 IEEE . Institute of Electrical and Electronics Engineers, no. 2 (vol. 104), pp. 444-466.
[4] S. Nemati, A. Holder, F. Razmi, M. Stanley, G. Clifford and T. Buchman, “An In- terpretable Machine Learning Model for Accurate Prediction of Sepsis in the ICU”, Critical Care Medicine, vol. 46, no. 4, pp. 547–553 ,2018.
[5] A.Vellido, F. Schleif and M. Biehl, ”Advances in machine learning and computational intelligence”, Neurocomputing, vol. 72, no. 7-9, pp. 1377-1378, 2009.
[6] K. Barajas, R. Akella, “Prediction of Pysiological Subsystem Failure and its Impact in the prediction of Patient Mortality”, In the Proceedings of IEEE, Internaltional Conference on Big Data, 2015.
[7] P. Thottakkara, T. Ozrazgat-Baslanti, B. Hupf, P. Rashidi, P. Pardalos, P. Momcilovic, A. Bihorac, “Application of Machine learning Techniques to High Dimensional Clinical to Forecast Postoperative Complications”, PLOS ONE, vol. 11, no. 5, 2016.
[8] S. Bhattacharya, V. Rajan, H. Shrivastava, “ICU Mortality Prediction: Classification Algorithm for Imbalanced Datasets”,In the Proceedings of the 31st AAAI Conference on Artificial Intelligence (AAAI-17), pp. 1228-1294, 2017.
[9] Soni, HK., “Machine Learning – A New Paradigm of AI”, International Journal of Scientific Rsearch in Network Security and Communication, pp. 31-32,2019
[10] Ghosh, S. & Waheed, Sajjad, “Strategic Analysis of classification algorithms for liver disease diagnosis”, Journal of Science Technology and Environment Informatics,pp. 361-370, 2017.
Downloads
Published
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors contributing to this journal agree to publish their articles under the Creative Commons Attribution 4.0 International License, allowing third parties to share their work (copy, distribute, transmit) and to adapt it, under the condition that the authors are given credit and that in the event of reuse or distribution, the terms of this license are made clear.
