Recent Submissions

Thumbnail Image
PublicationOpen Access
A machine learning approach for marginal fulfillment cost estimation in last mile delivery
(Elsevier, 2025-07-01) Nalbant, Ali; Yıldız, Barış; Graduate School of Sciences and Engineering; Department of Industrial Engineering; GRADUATE SCHOOL OF SCIENCES AND ENGINEERING; College of Engineering
Determining marginal fulfillment costs (MFC) is crucial for effective decision-making in online grocery retail, a sector struggling with small profit margins and arduous service requirements of attended home deliveries. Paramount to improving operational efficiency, e-grocers need accurate real-time MFC estimations to optimize their service offers and prices for online customers. Traditional methods for estimating MFC are either too slow for online decision-making or inaccurate. This paper introduces a novel machine learning (ML) approach that provides fast and accurate MFC estimations with the help of carefully engineered features (predictors) that can capture complex routing dynamics. Experiments with real-world data demonstrate the superiority of the proposed approach over state-of-the-art MFC estimation methods. Our analysis of more than 2000 potential predictors, from which 20 are curated for practical applicability, reveals critical insights into the use of network-level, neighborhood-based, and node-level features in capturing complex VRP dynamics to develop ML-based approaches to address problems that arise in different transportation applications.
Thumbnail Image
PublicationOpen Access
Polyurethane synthesis revisited: effect of solvent, stoichiometry, and temperature on the reaction of MDI with polyether glycols
(Elsevier, 2025-05-09) Yılgör, İskender; Yılgör, Emel; Yıldırım, Armen; Department of Chemistry; KUYTAM (Koç University Surface Science and Technology Center); Graduate School of Sciences and Engineering; College of Sciences; GRADUATE SCHOOL OF SCIENCES AND ENGINEERING; Graduate School of Sciences and Engineering; Research Center
Thermoplastic polyurethanes (TPU) are one of the most widely investigated polymeric systems due to their interesting structure-morphology-property behavior. They also find broad range of applications in various fields. Global TPU market is projected to grow about 7.3 % annually from $2.30 billion in 2021 to $3.80 billion in 2028. 4,4′-Diphenylmethane diisocyanate (MDI) is the most widely used diisocyanate for the preparation of TPUs both in academia and industry. When TPU synthesis is carried out in solution, a polar aprotic solvent is necessary to obtain high molecular weight polymers. Most preferred solvents for TPU synthesis are high boiling, polar, aprotic solvents, such as dimethylacetamide (DMAC), dimethylformamide (DMF), dimethyl sulfoxide (DMSO) and N-methyl pyrrolidone (NMP). When MDI is used as the diisocyanate, depending on the solvent used and reaction temperature, extensive side reactions may be observed, which consume excess diisocyanate and affect reaction stoichiometry. Side reactions also strongly influence TPU structure, topology, microphase morphology, and properties. In this study influence of the solvent, initial [NCO]/[OH] stoichiometry and reaction temperature on the rate of isocyanate consumption and kinetics of the reactions between MDI and poly(tetramethylene oxide) glycol (PTMO) were investigated. Catalytic effect of DMF even at reactions conducted at room temperature were observed, resulting in significant excess MDI consumption due to extensive side reactions. During prepolymer formation in [MDI]/[PTMO] = 2.0 system at 50 °C, side reactions were minimized or eliminated by using THF/DMF or toluene/DMF (90/10 by volume) solvent mixtures.
Thumbnail Image
PublicationOpen Access
Simple and green process for silk fibroin production by water degumming
(American Chemical Society, 2025-01-05) Atay, İpek; Yağcı, Mustafa Barış; Sürme, Saliha; Kavaklı, İbrahim Halil; Yılgör, Emel; Yılgör, İskender; KUYTAM (Koç University Surface Science and Technology Center); Department of Chemistry; Department of Chemical and Biological Engineering; Department of Molecular Biology and Genetics; College of Sciences; College of Engineering; Research Center
Silk fibroin (SF), a natural polymer with very desirable physicochemical and biological properties, is an ideal material for crafting biocompatible scaffolds in tissue engineering. However, conventional methods for removing the sericin layer and dissolving SF often involve environmentally harmful reagents and processes, requiring extensive dialysis procedures to purify the fibers produced. Such processes may also damage the surface and bulk properties of the SF produced. Here, we report a simple, green water degumming method, in which almost complete sericin removal of 30% by weight is achieved in 6 h in boiling water. The SF produced is easily dissolved in formic acid/orthophosphoric acid (90/10, 85/15, and 70/30) mixtures, eliminating the need for salts like LiBr and CaCl2 followed by dialysis and freeze-drying, thus simplifying the process significantly. Additionally, our findings demonstrate significantly enhanced cell viability in electrospun poly(lactic acid)/SF blends. Overall, SF production via water degumming offers an eco-friendly pathway for generating bioactive scaffolds in tissue engineering applications.
Placeholder
Publication
Natural history of PIRADS-2 lesions on serial multiparametric magnetic resonance imaging: real-life data from an Academic Center
(Elsevier, 2025) Esen, Barış; Gürses, Bengi; Kordan, Yakup; Kiremit, Murat Can; Tilki, Derya; Esen, Tarık; School of Medicine; SCHOOL OF MEDICINE
Introduction/Background: The natural history of prostate imaging reporting and data system (PIRADS) score 2 lesions on serial mpMRIs is largely unknown. Herein, we aimed to evaluate the patients with PIRADS-2 index lesions by using serial mpMRI scans to reveal the rates of mpMRI upgrade in PIRADS score and prostate cancer (PCa) detection. Methods/Materials: All mpMRI scans with a PIRADS-2 index lesion from our mpMRI database were evaluated retrospectively. Data from 214 biopsy-na and imath;ve patients with a PIRADS-2 index lesion on the initial mpMRI who then underwent at least 1 follow-up mpMRI were reevaluated by an experienced uroradiologist and only those (n = 172) who had a PIRADS-2 index lesion on the initial mpMRI according to PIRADS v2.1 were included in the study. mpMRI progression was defined as the detection of any PIRADS >= 3 lesion at follow-up mpMRI. Histopathological results were evaluated in patients undergoing biopsy upon mpMRI progression. Results: A total of 172 patients with a mean age of 60.1 8.6 years were evaluated. The median PSA at baseline mpMRI was 4.7 (IQR;3.3-6.7) ng/dl. Overall mpMRI progression was detected in 54 patients (31.4%), 37 were upgraded to PIRADS-3, 16 to PIRADS-4, and one to PIRADS-5. Multivariate logistic regression analysis revealed that a PSA increase of >= 25% during follow-up was the only predictor of mpMRI upgrade (P = 0.019, OR: 2.384). 30 out of 54 patients underwent a prostate biopsy and PCa was detected in 15 patients;5 with ISUP grade 1, 10 with ISUP grade 2. Conclusions: Almost half of the patients with a PIRADS-2 index lesion were upgraded to PIRADS >= 3 when evaluated with serial mpMRI when a PSA increase of >= 25% was observed during follow-up. PCa was detected in half of the patients who underwent a biopsy. Serial mpMRI can be recommended when monitoring patients with elevating PSA >= 25%, a prostate biopsy can be considered upon a mpMRI progression. (c) 2024 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Placeholder
Publication
CBWO: a novel multi-objective load balancing technique for cloud computing
(Elsevier, 2025) Hayyolalam, Vahideh; Özkasap, Öznur; Department of Computer Engineering; Graduate School of Sciences and Engineering; College of Engineering; GRADUATE SCHOOL OF SCIENCES AND ENGINEERING
In cloud computing systems, the growing demand for diverse applications has led to challenges in resource allocation and workload distribution, resulting in increased energy consumption and computational costs. To address these challenges, we propose a novel load-balancing method, namely CBWO, that integrates Chaos theory with the Black Widow Optimization algorithm. Our approach is designed to optimize cloud computing environments by improving energy efficiency and resource utilization. We employ CloudSim for simulations, evaluating key performance metrics such as energy consumption, resource utilization, makespan, task completion time, and imbalance degree. The experimental results demonstrate the superiority of our method, achieving average improvements of 67.28% in makespan and 29.03% in energy consumption compared to existing solutions.