Kocaeli University Research Information System
Not a member yet
80615 research outputs found
Sort by
Crowdfunding as an E-Commerce Mechanism: A Deep Learning Approach to Predicting Success Using Reduced Generative AI Embeddings
No full textCrowdfunding platforms like Kickstarter have reshaped early-stage financing by allowing entrepreneurs to connect directly with potential supporters. As a fast-expanding part of digital commerce, crowdfunding offers significant opportunities but also substantial risks for both entrepreneurs and platform operators, making predictive analytics an essential capability. Although crowdfunding shares some operational features with traditional e-commerce, its mix of financial uncertainty, emotionally charged storytelling, and fast-evolving social interactions makes it a distinct and more challenging forecasting problem. Accurately predicting campaign outcomes is especially difficult because of the high-dimensionality and diversity of the underlying textual and behavioral data. These factors highlight the need for scalable, intelligent data science methods that can jointly exploit structured and unstructured information. To address these issues, this study proposes a novel AI-based predictive framework that integrates a Convolutional Block Attention Module (CBAM)-enhanced symmetric autoencoder for compressing high-dimensional Generative AI (GenAI) BERT embeddings with meta-heuristic feature selection and advanced classification models. The framework systematically couples attention-driven feature compression with optimization techniques—Genetic Algorithm (GA), Jaya, and Artificial Rabbit Optimization (ARO)—and then applies Long Short-Term Memory (LSTM) and Gradient Boosting Machine (GBM) classifiers. Experiments on a large-scale Kickstarter dataset demonstrate that the proposed approach attains 77.8% accuracy while reducing feature dimensionality by more than 95%, surpassing standard baseline methods. In addition to its technical merits, the study yields practical insights for platform managers and campaign creators, enabling more informed choices in campaign design, promotional tactics, and backer targeting. Overall, this work illustrates how advanced AI methodologies can strengthen predictive analytics in digital commerce, thereby enhancing the strategic impact and long-term sustainability of crowdfunding ecosystems
On some novel Boole’s type inequalities for conformable fractional integrals with their applications
No full textThe extension of classical numerical quadrature rules, such as Boole’s rule, into the fractional calculus framework is an essential step for enhancing computational methods in applied mathematics. However, a significant challenge has been the development of sharp error bounds that are applicable under minimal differentiability conditions, particularly for the rapidly evolving field of Conformable fractional integrals. This work addresses this gap by first establishing a novel Boole-type identity within the Conformable fractional paradigm. This foundational identity serves as the critical tool to derive new and refined inequalities specifically designed for convex functions. A primary objective and achievement of this research is the determination of improved error bounds that require only a single time differentiability, significantly broadening their applicability. Our results not only generalize existing theorems but also provide sharper estimates for numerical integration. To ensure the validity and applicability of our proposed results, we demonstrate a brief numerical investigation, which not only supports the theoretical findings but also illustrates their relevance to practical and computational problems. Moreover, the theoretical inequalities are then applied to the error analysis of quadrature formulas and to derive new relationships for Mittag-Leffler function and special means for real numbers. Our findings extend classical Boole’s rule to the fractional setting and provide tighter error bounds than already existed in the literature
Photocatalytic Removal of CO From Polluted Air by TiO2 and ZSM-5 Supported ZnCr2O4 Nanocatalysts—Catalyst Optimization by Response Surface Methodology
No full textThe photocatalytic oxidation of carbon monoxide (CO) in polluted air was investigated using ZnCr2O4 nanocatalysts supported on TiO2 and ZSM-5, synthesized via a sol–gel auto-combustion method. Comprehensive characterization including Fourier transform infrared (FTIR), scanning electron microscopy (SEM), x-ray photoelectron spectroscopy (XPS), photoluminescence spectroscopy (PL), x-ray diffraction (XRD), UV–vis diffuse reflectance spectroscopy (DRS), and Brunauer–Emmett–Teller (BET) analysis was conducted to elucidate the structural, optical, and surface properties. Due to the superior performance of ZnCr2O4/ZSM-5, subsequent investigations on CO removal were focused on this catalyst, and response surface methodology (RSM) was applied to optimize its synthesis parameters. The optimized catalyst, with a ZnCr2O4:ZSM-5 mass ratio of 0.27:0.73 and calcined at 700°C, achieved maximum removal efficiency at an operating temperature of 25–40°C, as confirmed by strong agreement between the experimental and predicted RSM results (R2 = 93.5). The findings highlight the dominant influence of support dosage on photocatalytic activity, while minimal water vapor was found to be critical for efficient hydroxyl radical formation. In contrast, high humidity reduced performance by inhibiting surface adsorption. Pareto analysis showed that the support dosage had the greatest influence on the efficiency of ZnCr2O4/ZSM-5. The band gap of ZnCr2O4and ZnCr2O4/ZSM-5 was determined to be 2.9 and 2.8 eV, respectively
Glycaemic control and complications in haemodialysis patients: The TURK-HEMODIAB Study
No full textBackground The most common cause of end-stage kidney disease is diabetes mellitus (DM). The most commonly used renal replacement therapy in Turkey and in many countries around the world is haemodialysis (HD). Glycaemia control is important in these populations. In this study we aimed to screen for glycaemic control and complications in a large population of diabetic HD patients in Turkey. Methods A total of 16 043 patients were screened in 253 dialysis centres in Turkey and 5038 diabetic HD patients were included in the study. At participating centres, patients' diabetes history, complications, medications, haemoglobin A1c (HbA1c) and other laboratory data were reviewed and recorded by nephrologists. Results The average age of the patients was 64.0 ± 11.2 years and 56% were male. The mean HbA1c was 7.4 ± 1.5%. Patients were divided into three groups according to the HbA1c level (8%). As the HbA1c levels increased, the mean systolic blood pressure and diastolic blood pressure increased significantly. In addition, as the HbA1c levels increased, the number of patients with coronary artery disease, patients undergoing coronary artery bypass graft surgery and the rate of patients with diabetic retinopathy and vision loss increased. Diabetic foot disease and amputation rates were also higher in the group with poor glycaemic control. The number of patients using intensive or mixed insulin was also higher in the group with high HbA1c levels. In ordinal logistic regression analysis, age significantly decreased and higher body mass index slightly increased the risk of a higher HbA1c. Also, the need for a diabetic diet was greater in those with high HbA1c levels. Conclusion Our study highlights that the target values for diabetic HD patients in Turkey are partially compatible with the 2022 Kidney Disease: Improving Global Outcomes guidelines for diabetes management. Nevertheless, more effort and teamwork are needed to improve patient outcomes
Exploring an independent association between rituximab-cyclophosphamide-doxorubicin-vincristine-prednisolone dose intensity and clinical outcomes in large B-cell lymphoma: Analyses of subsequent endpoints following a complete response in a real-world cohort of 1369 patients
No full textIntroduction: Rituximab-cyclophosphamide-doxorubicin-vincristine-prednisolone (R-CHOP) is a standard first-line treatment in large B-cell lymphoma (LBCL). Dose intensity is frequently reduced because of toxicity or concerns of tolerability. The true impact of such treatment modifications on the response rates is difficult to assess in nonrandomized comparisons because of complex and reciprocal interactions between various determinants of both variables. To avoid these confounding biases, this study was designed to analyze subsequent outcomes after a complete response is achieved with frontline therapy, aiming to uncover an independent link between R-CHOP dose-intensity and long-term outcomes in LBCL. Methods: Patients treated between 2012 and 2024 were included. Reduced dose-intensity was predefined as more than 20% dose reduction for any one of the drugs for two cycles or a cumulative delay of ≥21 days. The primary and secondary objectives were the duration of complete response and the cumulative incidence of relapse (CIR), respectively. Results: The LBCL cohort consisted of 1369 consecutive cases. For this study, 953 were eligible, with 728 and 225 patients in the dose-intense and reduced-intensity groups, respectively. Duration of complete response was significantly longer for the dose-intense group with a 3-year estimate of 87.8% vs 68.8% (hazard ratio: 0.39, p <.0001). The estimated 3-year cumulative incidence of relapse was 10.3% vs 21.4% (hazard ratio: 0.51, p =.004), favoring the dose-intense group. Reduced dose intensity was an independent risk factor on multivariate analysis for both endpoints. Conclusions: Focusing on long-term outcomes after CR, this study demonstrates an increased risk of relapse when R-CHOP intensity is reduced, providing novel evidence that complements the association between reduced treatment intensity and inferior LBCL outcomes
ONKOPLASTİK MEME CERRAHİSİ TEKNİKLERİNİN TEMELLERİ
No full textCopy Download .BUmberto Veronesi, pioneer of modern breast surgery, declared that ladies who are conscious about breast cancer and attend screening programs must be encouraged and must not be punished with heavy and unacceptable treatments. From yesterday to today, breast cancer surgery has changed from radical mastectomy to oncoplastic breast surgery (OPBS) with oncological and aesthetic targets. The philosophy of OBS integrates oncological safety with cosmetic principles to enhance the psychological and physical well-being of breast cancer patients. The core principles of the philosophy of oncoplastic surgery are represented below. The primary goal of OPBS is to remove the tumor with clear margins for oncological safety. OPBS embodies a patient-centered approach that combines effective oncologic treatment with a commitment to aesthetic preservation. Oncoplastic Techniques aim to maintain or restore the natural shape and symmetry of the breast. Patients are involved in decision-making, with a clear understanding of their surgical options, Success requires collaboration among surgeons, oncol-ogists, radiologists, pathologists, and plastic surgeons to create a unified treatment plan. Techniques focus on reducing surgical complications and preserving breast functionality. OPBS helps patients regain body confidence and self-esteem and increases Quality of Life. OPBS mixes oncological and plastic surgery principles, creating innovative procedures tailored to individual patients. Psychological counseling and patient education are integral to the care plan. The training of surgeons in both onco-logical and reconstructive techniques offers optimal care. In essence, the philosophy of oncoplastic surgery revolves around treating breast cancer not only as a disease to be cured but as a condition that requires comprehensive care to restore the patient's overall sense of identity, dignity, and quality of life. Surgeons and patients still debate which of the oncological results, cosmetic data, and changes in qual-ity of life taken into account for evaluating disease-free survival and local recurrence are important</div
Advances in Magnéli phase Ti4O7 materials for water and wastewater treatment: synthesis, fabrication, and future perspectives
No full textMagnéli phase titanium oxides constitute a series of electrically conductive ceramic materials that have gained significant interest for nearly seven decades. Magnéli Ti4O7 is the most investigated among the series for its unique blend of superior electrical conductivity and corrosion resistance. Such qualities of Ti4O7 have found useful applications in materials science, chemistry, physics, and environmental engineering. Ti4O7 materials are usually synthesized through the thermal reduction of TiO2 with hydrogen, carbon, metals, or metal hydrides under atmospheric and time control. Ti4O7 materials produced through carbothermal, metallothermic, and H2, exhibit significantly higher electrical conductivity than those synthesized via sol–gel and metal hydrides techniques. They are typically used as anodic electrodes or reactive electrochemical membranes (REMs) in advanced electrochemical water treatment processes. Ti4O7 anodes and REMs have demonstrated better pollutant removal efficiency and electrochemical stability than dimensionally stable anode (DSA) electrodes and are comparable to the expensive boron-doped diamond (BDD) anodes, making them cost-effective substitutes. However, they suffer mass and charge transfer resistance limitations attributed to a thick boundary layer, which significantly affects the generation of oxidative species and their electrocatalytic activity. Recent research is focused on developing Ti4O7 materials with enhanced catalytic and stability properties, through synthesis process optimization, nanostructuring, controlled defect engineering, doping, and composite fabrication
Generative AI in English-medium instruction: Perceptions, usage, and impact on academic performance and language proficiency
No full textThis study investigated the perceptions and use of Generative Artificial Intelligence (GenAI) among 387 social sciences students within an English-Medium Instruction (EMI) context at a Turkish university. Employing a quantitative cross-sectional survey design, the research examined how students' GenAI-related characteristics (usage time, experience, self-perceived competence) and modes of GenAI integration (complementary, substitutive, or hybrid) correlated with their academic performance, which was measured through grade point average (GPA) and English language proficiency. Additionally, open-ended responses regarding tool usage were subjected to quantitative content analysis to identify prevalence trends. Key findings indicated that language proficiency positively correlated with EMI academic performance and negatively correlated with substitutive GenAI use, which suggested that students with lower proficiency may use GenAI to compensate for linguistic challenges. Conversely, complementary GenAI use was positively associated with academic performance, which highlighted its benefits. Frequency analysis revealed that students predominantly used ChatGPT for translation, proofreading, and idea generation, and frequently used other GenAI assisted tools like Grammarly and Duolingo. These results offer crucial insights into GenAI's integration within EMI learning, which inform the design of effective GenAI-supported pedagogical practices
