Araştırma Çıktıları | TR-Dizin | WoS | Scopus | PubMed
Permanent URI for this communityhttps://hdl.handle.net/11727/4806
Browse
5 results
Search Results
Item A deep learning approach for sepsis monitoring via severity score estimation(2021) Asuroglu, Tunc; Ogul, Hasan; 33157471Background and objective: Sepsis occurs in response to an infection in the body and can progress to a fatal stage. Detection and monitoring of sepsis require multi-step analysis, which is time-consuming, costly and requires medically trained personnel. A metric called Sequential Organ Failure Assessment (SOFA) score is used to determine the severity of sepsis. This score depends heavily on laboratory measurements. In this study, we offer a computational solution for quantitatively monitoring sepsis symptoms and organ systems state without laboratory test. To this end, we propose to employ a regression-based analysis by using only seven vital signs that can be acquired from bedside in Intensive Care Unit (ICU) to predict the exact value of SOFA score of patients before sepsis occurrence. Methods: A model called Deep SOFA-Sepsis Prediction Algorithm (DSPA) is introduced. In this model, we combined Convolutional Neural Networks (CNN) features with Random Forest (RF) algorithm to predict SOFA scores of sepsis patients. A subset of Medical Information Mart in Intensive Care (MIMIC) III dataset is used in experiments. 5154 samples are extracted as input. Ten-fold cross validation test are carried out for experiments. Results: We demonstrated that our model has achieved a Correlation Coefficient (CC) of 0.863, a Mean Absolute Error (MAE) of 0.659, a Root Mean Square Error (RMSE) of 1.23 for predictions at sepsis onset. The accuracies of SOFA score predictions for 6 hours before sepsis onset were 0.842, 0.697, and 1.308, in terms of CC, MAE and RMSE, respectively. Our model outperformed traditional machine learning and deep learning models in regression analysis. We also evaluated our model's prediction performance for identifying sepsis patients in a binary classification setup. Our model achieved up to 0.982 AUC (Area Under Curve) for sepsis onset and 0.972 AUC for 6 hours before sepsis, which are higher than those reported by previous studies. Conclusions: By utilizing SOFA scores, our framework facilitates the prognose of sepsis and infected organ systems state. While previous studies focused only on predicting presence of sepsis, our model aims at providing a prognosis solution for sepsis. SOFA score estimation process in ICU depends on laboratory environment. This dependence causes delays in treating patients, which in turn may increase the risk of complications. By using easily accessible non-invasive vital signs that are routinely collected in ICU, our framework can eliminate this delay. We believe that the estimation of the SOFA score will also help health professionals to monitor organ states. (C) 2020 Elsevier B.V. All rights reserved.Item Predicting Bank Return on Equity (ROE) using Neural Networks(2021) Balci, Tolgay; Ogul, HasanMeasuring the performance and profitability of the banking sector, which is the most important part of a country's financial system, is always important. Thanks to the performance measurement, banks can understand the competitive situation, their potential to grow, and the risk, and be more successful in sustaining their lives. This study is considered all state deposit money banks in Turkey. In the literature, using of artificial neural networks (ANN) in banking performance evaluation is rarely studied. Therefore, this paper aims to examine the possibility of ANN utilization for predicting return on equity of Turkey State Deposit Money Banks. The paper compares the accuracy percentages of optimization algorithms of ANN using eleven years quarterly data of six exogenous variables and eight endogenous variables as independent variables and the average return on equity from quarterly of all Turkey state deposit money banks as dependent variable. Given a number of recorded financial parameters, the task is to predict banks' performances using ANN computation methods and to compare prediction results with real results. To evaluate these methods, we built a data set from Banking Regulation and Supervison of Agency, The Banks Association of Turkey and banks' quarterly financial reports. According to all experimental results in optimization models were estimated with above % 80 accuracy. It is determined that the best optimization model is different for each bank.Item Prediction of Protein Metal Binding Sites Using Deep Neural Networks(2019) Haberal, Ismail; Ogul, Hasan; 0000-0002-8647-4295; 30977960Metals have crucial roles for many physiological, pathological and diagnostic processes. Metal binding proteins or metalloproteins are important for metabolism functions. The proteins that reach the three-dimensional structure by folding show which vital function is fulfilled. The prediction of metal-binding in proteins will be considered as a step-in function assignment for new proteins, which helps to obtain functional proteins in genomic studies, is critical to protein function annotation and drug discovery. Computational predictions made by using machine learning methods from the data obtained from amino acid sequences are widely used in the protein metal-binding and various bioinformatics fields. In this work, we present three different deep learning architectures for prediction of metal-binding of Histidines (HIS) and Cysteines (CYS) amino acids. These architectures are as follows: 2D Convolutional Neural Network, Long-Short Term Memory and Recurrent Neural Network. Their comparison is carried out on the three different sets of attributes derived from a public dataset of protein sequences. These three sets of features extracted from the protein sequence were obtained using the PAM scoring matrix, protein composition server, and binary representation methods. The results show that a better performance for prediction of protein metal- binding sites is obtained through Convolutional Neural Network architecture.Item A New Approach for Predicting the Value of Gene Expression: Two-way Collaborative Filtering(2019) Bayrak, Tuncay; Ogul, Hasan; 0000-0001-6826-4350; AAE-3731-2020; AAE-3731-2020Background: Predicting the value of gene expression in a given condition is a challenging topic in computational systems biology. Only a limited number of studies in this area have provided solutions to predict the expression in a particular pattern, whether or not it can be done effectively. However, the value of expression for the measurement is usually needed for further meta-data analysis. Methods: Because the problem is considered as a regression task where a feature representation of the gene under consideration is fed into a trained model to predict a continuous variable that refers to its exact expression level, we introduced a novel feature representation scheme to support work on such a task based on two-way collaborative filtering. At this point, our main argument is that the expressions of other genes in the current condition are as important as the expression of the current gene in other conditions. For regression analysis, linear regression and a recently popularized method, called Relevance Vector Machine (RVM), are used. Pearson and Spearman correlation coefficients and Root Mean Squared Error are used for evaluation. The effects of regression model type, RVM kernel functions, and parameters have been analysed in our study in a gene expression profiling data comprising a set of prostate cancer samples. Results: According to the findings of this study, in addition to promising results from the experimental studies, integrating data from another disease type, such as colon cancer in our case, can significantly improve the prediction performance of the regression model. Conclusion: The results also showed that the performed new feature representation approach and RVM regression model are promising for many machine learning problems in microarray and high throughput sequencing analysis.Item On computer-aided prognosis of septic shock from vital signs(2019) Ogul, Hasan; Baldominos, Alejandro; Asuroglu, Tunc; Colomo-Palacios, Ricardo; AAC-7834-2020Sepsis is a life-threatening condition due to the reaction to an infection. With certain changes in circulatory system, sepsis may progress to septic shock if it is left untreated. Therefore, early prognosis of septic shock may facilitate implementing correct treatment and prevent more serious complications. In this study, we assess the feasibility of applying a computer-aided prognosis system for septic shock. The system is envisaged as a tool to predict septic shock at the time of sepsis onset using only vital signs which are collected routinely in intensive care units (ICUs). To this end, we evaluate the performances of computational methods that take the sequence of vital signs acquired until sepsis onset as input and report the possibility of progressing to a septic shock before any further clinical analysis is performed. Results show that an adaptation of multivariate dynamic time warping can reveal higher accuracy than other known time-series classification methods on a new dataset built from a public ICU database. We argue that the use of computational intelligence methods can promote computer-aided prognosis of septic shock in hospitalized environment to a certain degree.