Using artificial intelligence algorithms, researchers at the University's department of Information Technology Engineering developed a prognostic model for cardiac arrest in patients with sepsis. The proposed model predicts its occurrence with a sensitivity value above 70% in the period of 30 hours before the cardiac arrest.
Sepsis-associated cardiac arrest is a common issue with the low survival rate. Early prediction of cardiac arrest can provide the time required for intervening and preventing its onset in order to reduce mortality. Several studies have been conducted to predict cardiac arrest using machine learning. However, no previous research has used machine learning for predicting cardiac arrest in adult sepsis patients. Moreover, the potential of some techniques, including ensemble algorithms, has not yet been addressed in improving the prediction outcomes. It is required to find methods for generating high-performance predictions with sufficient time lapse before the arrest. In this regard, various variables and parameters should also been examined.
Dr. Samaneh Layeghian, whose research was conducted in the form of her PhD dissertation in the field of information technology engineering (information systems), stated: "Because physicians cannot continuously monitor the risk of cardiac arrest for all patients under care, automation of collection and analysis health data as well as the necessary warnings to the patient and the doctor can be a big step in reducing mortality and costs. She added: "In this study, using artificial intelligence algorithms, we developed a prognostic model for cardiac arrest for patients with sepsis."
Layeghian explained: In this regard, 30 hours of clinical data of sepsis patients were extracted from MIMIC III database (79 cases of cardiac arrest, 4532 normal records). Various machine learning models were trained (models for six time groups) with a systematic approach on these three datasets. The models included classical techniques (SVM, decision tree, logistic regression, KNN, GaussianNB) and ensemble methods (gradient Boosting, XGBoost, random forest, balanced bagging classifier and stacking). Proper solutions were proposed to address the challenges of missing values, imbalanced classes of data and irregularity of time series. Finally, the use of deep learning method produced better results.
She added: "We illustrated that machine learning techniques, especially ensemble algorithms, have high potentials to be used in prognostic systems for sepsis patients. Comparing the output of this model with the results of the two standard warning systems Apache II and MEWS showed that the proposed model has a significant improvement in the evaluation criteria, over existing standard systems. In this study, the effect of the dynamics of vital signs time series, as a predictor for predicting cardiac arrest, was also tested with different approaches. Time series analysis to predict cardiac arrest one hour before the event, produced a sensitivity of 77%. According to the results, the time series dynamics of vital signs are of great importance in the prediction of cardiac arrest incidence in sepsis patients.?
In the next step, in order to make the data collection and analysis operations intelligent, we used new technologies to design a high-level view of an "IOT" architecture to monitor real-time of patients in intensive care units. The architecture provides the proposed intelligent forecasting model as part of its services and uses "fog" technology to accelerate early processing operations.
This research was conducted in the form of Samaneh Layeghian's doctoral dissertation under the guidance of Dr. Mohammad Mehdi Sepehri, academic member of Industrial and Systems Engineering.