No abstract

Sepsis is a major public and global health concern. Every hour of delay in detecting sepsis significantly increases the risk of death, highlighting the importance of accurately predicting sepsis in a timely manner. A growing body of literature has examined developing new or improving the existing machine learning (ML) approaches for timely and accurate predictions of sepsis. This study contributes to this literature by providing clear insights regarding the role of the recency and adequacy of historical information in predicting sepsis using ML. To this end, we implemented a deep learning model using a bidirectional long short-term memory (BiLSTM) algorithm and compared it with six other ML algorithms based on numerous combinations of the prediction horizons (to capture information recency) and observation windows (to capture information adequacy) using different measures of predictive performance. Our results indicated that the BiLSTM algorithm outperforms all other ML algorithms and provides a great separability of the predicted risk of sepsis among septic versus non-septic patients. Moreover, decreasing the prediction horizon (in favor of information recency) always boosts the predictive performance; however, the impact of expanding the observation window (in favor of information adequacy) depends on the prediction horizon and the purpose of prediction. More specifically, when the prediction is responsive to the positive label (i.e., Sepsis), increasing historical data improves the predictive performance when the prediction horizon is short-moderate.

Occasionally, surgeons do need various types of information to be available rapidly, efficiently and safely during surgical procedures. Meanwhile, they need to free up hands throughout the surgery to necessarily access the mouse to control any application in the sterility mode. In addition, they are required to record audio as well as video files, and enter and save some data. This is an attempt to develop a comprehensive operating room information system called "Medinav" to tackle all mentioned issues. An integrated and comprehensive operating room information system is introduced to be compatible with Health Level 7 (HL7) and digital imaging and communications in medicine (DICOM). DICOM is a standard for handling, storing, printing, and transmitting information in medical imaging. Besides, a natural user interface (NUI) is designed specifically for operating rooms where touch-less interactions with finger and hand tracking are in use. Further, the system could both record procedural data automatically, and view acquired information from multiple perspectives graphically. A prototype system is tested in a live operating room environment at an Iranian teaching hospital. There are also contextual interviews and usability satisfaction questionnaires conducted with the "MediNav" system to investigate how useful the proposed system could be. The results reveal that integration of these systems into a complete solution is the key to not only stream up data and workflow but maximize surgical team usefulness as well. It is now possible to comprehensively collect and visualize medical information, and access a management tool with a touch-less NUI in a rather quick, practical, and harmless manner.

No abstract

In this paper, a fuzzy expert system based on adaptive neuro-fuzzy inference system (ANFIS) is introduced to assess the mortality after coronary bypass surgery. In preprocessing phase, the attributes were reduced using a univariant analysis in order to make the classifier system more effective. Prognostic factors with a p-value of less than 0.05 in chi-square or t-student analysis were given to inputs ANFIS classifier. The correct diagnosis performance of the proposed fuzzy system was calculated in 824 samples. To demonstrate the usefulness of the proposed system, the study compared the performance of fuzzy system based on ANFIS method through the binary logistic regression with the same attributes. The experimental results showed that the fuzzy model (accuracy: 96.4%; sensitivity: 66.6%; specificity: 97.2%; and area under receiver operating characteristic curve: 0.82) consistently outperformed the logistic regression (accuracy: 89.4%; sensitivity: 47.6%; specificity: 89.4%; and area under receiver operating characteristic curve: 0.62). The obtained classification accuracy of fuzzy expert system was very promising with regard to the traditional statistical methods to predict mortality after coronary bypass surgery such as binary logistic regression model.

This study examines long-term care (LTC) discharge planning among older delayed discharge patients. While awaiting placements in alternate care such as LTC, these patients occupy hospital beds despite not requiring an intensive level of care. This study proposes a novel discharge decision model based on the Markov decision process (MDP) framework, which incorporates predictions regarding the patients' health trajectory and the associated hospital costs. Our machine learning (ML)-based predictive analytics allow for considering heterogeneous health transitions, hence personalized decision making, leading to valuable information for reducing hospital costs. We also develop data-driven cost functions using patient characteristics to estimate the personlevel costs associated with the decisions in the optimization model, that is, whether or not to discharge a patient to LTC. The data analyses and cost estimations are based on large historical data collected over 13 years in Ontario, Canada. To solve the resulting high-dimensional MDP models, we develop an index policy, where each patient's index value is calculated using their health complexity (comorbidity), sex, age, and acute length of stay in the hospital. Using extensive numerical experiments, we illustrate the superior performance of the proposed index policy against some benchmarking policies and demonstrate the significance of predictive information in optimizing discharge decisions. Our results also indicate that the value of predictive information increases with LTC bed availability and decreases with hospital capacity. We also demonstrate that with the anticipated exacerbating mismatch between supply and demand, targeted prediction-driven discharge policies, such as the proposed index policy, become even more critical. K E Y W O R D Sdelayed discharge, discharge planning, Markov decision process, predictive analytics, value of information 1

Cardiac events could be taken into account as the leading causes of death throughout the globe. Such events also trigger an undesirable increase in what treatment procedures cost. Despite the giant leaps in technological development in heart surgery, coronary surgery still carries the high risk of the mortality. Besides, there is still a long way ahead to accurately predict and assess the mortality risk. This study is an attempt to develop an expert system for the risk assessment of mortality following the cardiac surgery. The developed system involves three main steps. In the first step, a filtering feature selection method is applied to select the best features. In the second step, an ad hoc data-driven method is utilized to generate the preliminary fuzzy inference system. Finally, a hybrid optimization method is presented to select the optimum subset of the rules. The study relies on 1,811 samples to evaluate the diagnosis performance of the proposed system. The obtained classification accuracy is very promising with regard to other benchmark classification methods including binary logistic regression (LR) and multilayer perceptron neural network (MLP) with the same attributes. The developed system leads to 100% sensitivity and 84.7% specificity, while LR and MLP methods statistically come up with lower figures (65, 78.6 and 65%, 75.8%), respectively. Now, a fuzzy supportive tool can be potentially taken as an alternative for the current mortality risk assessment system that are applied in coronary surgeries, and are chiefly based on crisp database.