Ufuk University

This study examined the effectiveness of TheraBuddy, a web-based interactive visual novel, in enhancing bullying bystander intervention and socioemotional health among K-12 students. Using a four-wave controlled time-series design, 123 students aged 10-16 were randomly assigned to intervention (n = 62) or control (n = 61) groups. Participants completed validated measures at baseline and three-monthly follow-ups. The experimental group showed significantly greater gains in bystander intervention than controls, with effects moderated by bullying role and engagement level; socioemotional health effects were non-significant. Findings highlight the potential of interactive visual novels to promote prosocial action against bullying and inform scalable, school-based prevention strategies.

This study analyzes forest fires in Turkey between 2012 and 2022 using machine learning and artificial neural network methods, considering their timing, causes, burned area size, and spatial distribution. Official and publicly available data from the Turkish Orman Genel M & uuml;d & uuml;rl & uuml;& gbreve;& uuml; were used as the dataset. Decision Tree, Random Forest, Support Vector Machines, Multilayer Perceptron, and Long Short-Term Memory models were applied and comparatively evaluated. The performance of the classification models was evaluated using accuracy and F1 score, while temporal predictions were assessed using the Root Mean Square Error metric. According to the results, the Multilayer Perceptron model showed the highest success in predicting fire causes and burned area size classes (accuracy = 86%, F1 = 0.83). The Random Forest model similarly demonstrated strong performance with 85% accuracy and an F1 score of 0.82. The Long Short-Term Memory model, used for temporal predictions, achieved the lowest error value (Root Mean Square Error = 0.12) in estimating the annual and monthly number of fires and the amount of burned area, successfully capturing the variability, especially in years with excessive fires. Overall, the findings show that the Long Short-Term Memory model is superior in modeling temporal dependencies, while the Random Forest and Multilayer Perceptron models provide reliable results in classification problems. In conclusion, machine learning and artificial neural networks can be effectively used in the development of early warning systems, risk mapping, and decision support mechanisms for forest fires.

Triple-negative breast cancer (TNBC) is a distinct molecular subtype of breast cancer, characterized by high mortality and metastasis. Indoleamine 2,3-dioxygenase (IDO) is a highly expressed enzyme in cancer cells that contributes to immunosuppression. The IDO inhibition has been widely used in cancer immunotherapy. In this study, we evaluated the efficacy of indoximod, an IDO inhibitor, in combination with melatonin on TNBC tumor progression and metastasis in vitro and in vivo. Indoximod reduced the viability of 4T1 cells by inducing apoptosis and G2/M cell cycle arrest. Moreover, the combination of melatonin and TNF-α increased indoximod-mediated cell death. In vivo, coadministration of indoximod and melatonin resulted in enhanced antiproliferative and antimetastatic activities in the TNBC model. Tumor generation-induced neutrophil counts, TNF-α, IL-1β, and IL-10 levels were significantly reduced in mice receiving indoximod and melatonin in combination. Importantly, the accumulation of myeloid-derived suppressor cells (MDSCs), a key driver of tumor progression and metastasis, decreased in the primary tumor, metastatic liver, and lung of mice treated with indoximod and melatonin combination. Our findings revealed that melatonin enhances the immunotherapeutic efficacy of indoximod in TNBC.

This study delves into the intricacies of automatic lip reading (ALR) in the Turkish language, employing a deep learning model that integrates convolutional neural networks (CNNs) and long short-term memory (LSTM) units. By analyzing a comprehensive dataset comprising 111 words and 113 sentences, encompassing 67,080 video samples, the research identifies several key challenges inherent in Turkish ALR. The introduction of a novel "Sentences with Derived Words" (SDW) dataset underscores the impact of Turkish agglutinative morphology on ALR performance. Through rigorous analysis of commonly misclassified words, word lengths, phonetic resemblances, and consonant–vowel interactions, the study reveals that morphological diversity, phonetic similarities, and word lengths significantly influence model accuracy. Words containing bilabial consonants exhibit higher recognition rates, whereas shorter words and those with similar structures are more prone to misclassification. The SDW dataset, in particular, highlights the challenges posed by derived words, as evidenced by a notable decrease in word recognition rate. This work not only identifies critical limitations but also proposes recommendations for improving ALR models in Turkish, emphasizing the incorporation of phonetic features and the enhancement of dataset diversity. The findings offer profound insights into advancing ALR technologies for diverse applications, including assistive communication, forensic analysis, and human–computer interaction.

Team leadership training is essential alongside with technical training for effective resuscitation management. Addressing this gap, we developed a novel simulation system leveraging Large Language Models (LLMs) to create Artificial Intelligence (AI) agents simulating team members in Advanced Cardiovascular Life Support (ACLS) scenarios. This pilot study aimed to to develop a novel LLM-based ACLS simulation training platform and evaluate its performance in simulated resuscitation scenarios on established protocols. Using the Claude 3.5 Sonnet API, we designed a simulation system with four AI agents assigned specific roles as healthcare staff within an ACLS team. Each agent strictly followed the 2020 American Heart Association (AHA) ACLS guidelines while interacting with an ACLS certified emergency medicine specialist user. The ten patient scenario transcripts were evaluated with three blinded emergency medicine specialists whether all the recommended steps are completed. Inter-rater reliability was assessed using Kendall’s W and Krippendorff’s Alpha statistics to evaluate agreement both within raters and the model. AI agents consistently adhered to the AHA 2020 ACLS algorithm across scenarios, with a high inter-rater reliability (Kendall’s W > 0.75). Krippendorff’s Alpha values for agreement ranged from substantial (0.84) to almost perfect (0.99), indicating robust compliance with guidelines and effective simulation of resuscitation responses. This study highlights the potential of LLM-powered simulations as an adjunct to traditional resuscitation training. The system effectively supported team leadership training by providing consistent and guideline-compliant responses. While the results are promising, further research with larger participant samples is necessary to evaluate the long-term educational impact and scalability of such systems.