Background: Anemia due to malnutrition may develop as a result of iron, folate and vitamin B12 deficiencies. This situation poses a higher risk of morbidity and mortality in the geriatric population than in other age groups. Therefore, early diagnosis of anemia and early initiation of treatment is very important. This study aims to predict the diagnosis of anemia with using machine learning (ML) methods in geriatric patients followed in an outpatient clinic.Methods: In line with the purpose of the study, anemia classification was made by analysing patients' hemogram and biochemistry blood values and medical data such as malnutrition, physical and cognitive activity scores with ML methods.Results: In our data set consisting of 438 patient observations, the most successful ML algorithm was the J48 algorithm with 97.77% accuracy. In the continuation of the study, the predictive performance of anemia was investigated by excluding blood values and selecting only attributes consisting of malnutrition and physical activity scores. In this case, the most successful prediction was obtained with the Random Forest algorithm with 85.39% accuracy.Conclusions: The study showed that anemia can be predicted with high accuracy in geriatric patients without hemogram data. Additionally, our geriatric data set was shared with researchers for future research. Thus, it has contributed to the literature by opening a new path for studies on subjects such as comparing classification performances with new methodologies or predicting different diseases in geriatric patients.

Probability theory is a branch of science that statistically analyzes random events. Thanks to this branch of science, machine learning techniques are used inferences for the prediction or recommendation system. One of the statistical methods at the forefront of these techniques is Bayesian theory. Bayes is a simple mathematical formula used to calculate conditional probabilities and obtain the best estimates. The two most important parts of the formula are the concepts of a priori probability and posterior/conditional probability. In a priori probability, the most rational assessment of the probability of an outcome is made based on the available data, while in posterior probability, the probability of the event occurring is calculated after considering all evidence or data. In this study, a new mathematical model is presented to calculate the posterior probability variable of Bayesian theory more precisely. According to this new mathematical model, equal priority probabilities of some variables should be recalculated according to frequency. Calculations are applied to two nodes. The first of these two nodes is the node consisting of the existing data, and the second is the queried node. The positive frequency value will be applied when the variables consisting of existing data and having the same a priori probabilities are found at the questioned node, and negative frequency value will be applied for the other variables. Thus, while calculating a standard probability value according to Bayesian Theory, frequency-based values are taken into account with the help of the newly created mathematical model. With the help of these frequencies, the modification of the system reveals more precise results according to these two basic principles. The results obtained were tested with the cross validation method and high accuracy rates were determined.

While the dark web is designed to protect user privacy, it is also susceptible to misuse by malicious actors. Therefore, the effective classification of dark web (Tor and NonTor) traffic is of significant importance in cybersecurity. Findings from the literature reveal that time series-based methods are rarely employed in the detection of Tor network traffic. In this study, the performance of a long short-term memory (LSTM)-based deep learning model—commonly used in time series analysis—is investigated for the classification of Tor and NonTor network traffic alongside traditional machine learning techniques. This approach, which is supported by existing research findings, is further implemented within a graphical user interface (GUI) developed in the MATLAB environment. This interface allows users to upload datasets, apply data filtering, and perform detection using the model of their choice. The proposed system’s accuracy and overall performance demonstrate successful outcomes compared with results reported in the literature. The scientific contribution of this study lies in the development of a time series-based, GUI-supported application that presents a comparative evaluation of LSTM and conventional ML methods. Furthermore, the system provides both theoretical and practical advancements for network traffic analysis by integrating these models into a user-friendly interface.