Altınbaş University Institutional Repository
Not a member yet
5805 research outputs found
Sort by
An Investigation for Glaucoma and Segmentation Using Deep Network
No full textGlaucoma is a prevalent ocular condition that can lead to irreversible vision loss if left untreated in its early stages. Although optic imaging and machine learning have made significant advancements, there is still a gap in effectively and efficiently diagnosing glaucoma using fundus pictures. This research aims to address the existing gap in the field by introducing a novel deep learning approach for diagnosing glaucoma. We aim to develop a method that can precisely locate the optic disc in fundus photos, classify normal and glaucomatous photographs as healthy or glaucomatous with great precision, and effectively segment the optic disc region. To construct a multi-spatial network based on semantic links, it is crucial to merge a thorough collection of 12,000 fundus pictures obtained from seven distinct sources. The system we developed incorporates a convolutional neural network to extract features and perform classification tasks. This approach yields superior results compared to previous methods, achieving an accuracy of up to 94% on the ORIGA dataset. An automated diagnosis for glaucoma can have a substantial impact on therapy by enabling early intervention and improving outcomes
Machine Learning Services by Homomorphic Encryption to Enhancing Privacy
No full textMachine Learning (ML) is a crucial domain within data science, fostering advancements across several industries. Nonetheless, the rising threat of detrimental attacks on machine learning models presents significant privacy issues that may impede their extensive use. Privacy-Preserving Machine Learning (PPML) approaches, such as Homomorphic Encryption (HE), have been developed to mitigate these risks and safeguard sensitive data. Notwithstanding its potential, conventional HE encounters inefficiencies, especially in applications requiring great scalability. This study introduces a unique methodology, Hybrid Homomorphic Encryption (HHE), which integrates symmetric cryptography with homomorphic encryption to rectify these inefficiencies. We present the Guard ML framework, engineered for end devices, to enable encrypted data categorization while safeguarding the privacy of both the input data and the machine learning models. Employing a case study on heart disease categorization through sensitive ECG data, our technique demonstrates the practical use of HHE. Our methodology is viable because, although there is a minor reduction in accuracy relative to unencrypted inference, both analysts and end devices have minimal communication and computational costs. Our research establishes a foundation for a future of machine learning that prioritizes safety and privacy, particularly on resource-constrained end devices, by the successful integration of HHE into PPML
MPA + SVM: An Active Feature Selection Approach in High-Dimensional Data Sets
No full textAbstract: Picking a subset of possible features is a vital step in the data-mining procedure. The decisive goal of feature-selection is to determine the optimal number of superior characteristics to make best use of the presentation of the learning algorithm. However, this problem becomes increasingly challenging to resolve as the number of features in a data set increase. Therefore, to identify the optimal feature combinations, contemporary optimization techniques are used. Numerous optimization problems have been successfully resolved using the innovative metaheuristic known as the marine predators algorithm (MPA). Support vector machines (SVMs) are a crucial technique that is expertly applied to classification issues. In this work, the issue of feature picking in large dimensional data sets is solved by adjusting the MPA using the SVM as a classifier. In order to address the problem of feature selection in large dimensional data sets, the current study suggests MPA + SVM. Ten high-dimensional data sets got from the Arizona State University (ASU) source were employed to prove the usefulness of the planned method; the outcomes are likened with those of the additional six cutting-edge picking features algorithms. We compared the following algorithms: atom search optimization (ASO), satin bowerbird optimizer (SBO), emperor penguin optimizer (EPO), equilibrium optimizer (EO), monarch butterfly optimization (MBO), and sine cosine algorithm (SCA). The consequences confirm that the planned MPA+SVM approach outperformed several metaheuristic algorithms and presented a remarkable capability to pick the utmost weighty and optimum features. MPA + SVM yields the lowermost averaged error rates, minimal classification standard deviation (STD) values, and FS rates across all data sets. © Allerton Press, Inc. 2025
Design and Implementation of a LORAWAN Network-Based Low-Cost Photovoltaic (PV) Monitoring System
No full textSystems in response to the growing demand for effective monitoring frameworks in the renewable energy sector. The system incorporates sensors to record critical data, such as solar panel output, battery status, and ambient variables, addressing cost-effectiveness, scalability, and remote monitoring concerns in PV installations. Data transfer is supported via a LoRaWAN network, which provides long-range and low-power connectivity, making it ideal for remote and off-grid PV systems. The inclusion of the LoRaWAN technology improves system scalability and efficiency, providing a path for mass adoption in sustainable energy solutions. This low-cost design makes PV systems accessible to a wide range of users, including those in distant or economically challenged locations. The field testing results show that the system provides real-time data on PV system performance, enabling preventive maintenance and informed decision-making. The PZEM-004T V3 module and DC voltage sensor data provided important insights into the dynamic behavior of the monitored electrical system. The steady increase in active power consumption (Pac) from 933 to 2107 Watts implies an increase in the electrical power demand of the system. This study describes a realistic and efficient monitoring approach customized to the unique requirements of PV systems, contributing to the greater adoption of sustainable energy solutions using LoRaWAN technology
Ekstrakorporeal membran oksijenasyonundaki kritik hastalarda enfeksiyon ve antibiyotik paternleri
No full textBackground: This study aims to investigate the infection patterns and antibiotic utilization in critically ill patients receiving extracorporeal membrane oxygenation treatment. Methods: Between January 2019 and January 2024, a total of 165 patients (109 males, 56 females; median age: 58 years; range, 48 to 67 years) who were hospitalized for at least 24 h and underwent extracorporeal membrane oxygenation, and received ≥1 antibiotic treatment in the cardiovascular intensive care unit were retrospectively analyzed. Microbiological culture results, pathogen resistance patterns, antibiotics used, and their doses during extracorporeal membrane oxygenation were evaluated based on the literature and the Sanford Antimicrobial Guide database. Results: The median number of days spent on extracorporeal membrane oxygenation was 4 (range, 2 to 7) days. Klebsiella pneumoniae (28.8%) and Acinetobacter baumannii (21.1%) were frequently detected in culture results. The median number and duration of antibiotics were 2 (range, 1 to 3) and 2 (range, 1 to 4) days, respectively. Cephalosporins (39%) and penicillins (30%) were the most commonly used antibiotics. At least one antibiotic dose inappropriateness was detected in 56 (33.9%) patients. A total of 366 antibiotic administrations included 73 (19.9%) dose inappropriateness. Continuous renal replacement therapy, sepsis/septic shock, and extracorporeal membrane oxygenation duration >4 days were identified as risk factors increasing antibiotic inappropriateness (p<0.05). Conclusion: Our study results indicate that patients receiving extracorporeal membrane oxygenation frequently experience antibiotic resistance and the proliferation of Gram-negative bacteria. In our study, antibiotic dosing was inappropriate in approximately one-third of patients receiving extracorporeal membrane oxygenation. Based on these findings, adherence to the literature should be increased while selecting antibiotics and doses for patients
Kahramanmaraş Depremlerinden Etkilenmiş Bireylerle Çalışan Psikologlarda Beş Faktör Kişilik Özellikleri, Duygu Düzenleme ve İkincil Travmatik Stres Düzeyleri Arasındaki İlişki
No full textThis study examined the relationship between the Big Five personality traits, secondary traumatic stress levels, and emotion regulation strategies among psychologists working with individuals affected by the Kahramanmaraş earthquakes. The sample consisted of 201 psychologists. Data were collected via online methods using the “Demographic Information Form”, “Big Five-50 Personality Test”, “Secondary Traumatic Stress Inventory” and “Emotion Regulation Scale”. The findings indicated that higher levels of extraversion, agreeableness, conscientiousness, and emotional stability were associated with lower levels of secondary traumatic stress. Furthermore, extraversion and agreeableness showed a positive correlation with the suppression subdimension of emotion regulation, while openness to experience was positively related to both suppression and cognitive reappraisal strategies. Notably, changes in the emotional stability trait significantly predicted decreases in secondary traumatic stress, highlighting its important role as a predictor variable
Nano-Based Nasal Delivery of Biomacromolecules: A Myriad of Opportunities
No full textNasal cavity presents a highly suitable site for achieving therapeutic effects due to its relatively large, highly vascularized surface area for absorption and less harsh environment compared particularly with gastrointestinal tract and avoidance of first-pass metabolism. Moreover, it is easily accessible and provides a patient-friendly option for administration of medications. While the nasal route has been very well known for centuries, it has gained popularity in the last decades because of the myriad of opportunities that it offers to treat various conditions. However, mucociliary clearance, mucus layer, and enzyme activity in the nose limit the permeability of biomacromolecules through nasal membrane and, consecutively, utilization of nasal route for wider clinical conditions. Among several strategies explored to enhance nasal absorption, e.g., the use of penetration enhancers to improve nasal permeability, enzyme inhibitors to increase stability of biomacromolecules in nasal cavity, and incorporation of mucoadhesive agents to increase nasal residence time, nanocarriers show numerous advantages. Nano-based delivery systems have emerged as an attractive alternative to facilitate intranasal delivery of biomacromolecules through their inherited characteristics such as size, surface characteristics, and possibility for controlled release pattern. Biomacromolecules, such as therapeutic peptides, proteins, and nucleic acids, have poor nasal bioavailability because of their hydrophilicity and high molecular weight. Indeed, nasal delivery by nano-based systems has demonstrated the potential to increase their therapeutic efficacy while minimizing side effects. Moreover, nanocarriers make an attractive platform not only because of their generally non-toxic nature but also because it is possible to tailor their physicochemical characteristics to overcome several physiological barriers in the nose and promote targeted delivery. This chapter provides a comprehensive overview on nano-based nasal delivery of biomacromolecules. Firstly, the characteristics of biomacromolecules and their potential for clinical applications are highlighted. Secondly, factors in the nasal route that affect efficacy of nasally applied biomacromolecules are discussed considering their importance in the rational design of nanocarriers. The advantages and limitations of nasal route are also covered. Furthermore, various types of nano-based delivery systems and their formulation aspects are described, and the state-of-the-art knowledge about their use for nasal application of biomacromolecules is reviewed. The potential of nasal site for non-invasive delivery of biomacromolecules is handled out considering recent developments
A Dialectical Materialist Approach to Climate Justice: Akbelen Forest and Deştin Cement Factory Movements in Muğla
No full textThis article explores how a dialectical materialist approach enriches climate justice analysis by examining the contradictions between capital accumulation, labor, and nature. While conventional climate justice frameworks emphasize equitable outcomes and participatory governance, a dialectical materialist perspective delves deeper into the structural underpinnings of ecological degradation, linking environmental harm to capitalism's historical and material dynamics. By examining two case studies from Muğla, Türkiye, with empirical evidence collected from the field–the Akbelen movement against coal mining and the Deştin movement against cement factory construction–this article demonstrates how a dialectical approach reveals the systemic contradictions in climate justice analysis, historicizes ecological disparities, and questions the material basis of climate injustices. The study emphasizes the interconnectedness of environmental protection, class differentiation, and socio-economic emancipation. The findings contribute to environmental and climate justice scholarship by showcasing the analytical utility of dialectical materialism and shedding light on grassroots resistance in the Global South
A comprehensive risk evaluation method for sustainable building construction
Full text linkEffectively mitigating the risks associated with sustainable development is a key problem when building in developing nations. This method requires businesses to act following moral and ethical standards while balancing financial goals. Stakeholder consensus is critical for executing sustainable risk management, which strives to reduce risks while increasing possibilities. The research focuses on sustainability concerns in Iraqi construction projects, with the primary goal of identifying reasons causing project delays using a comprehensive framework and rigorous methodology. The initial stage of the approach is gathering data from several construction sites. To assess project risks, the weighted product approach was employed, which included factor scores from prior research to create structured questionnaires that were then used to study the influence of various aspects. To establish the significance of each aspect, the relative relevance index was used, and Delphi expert consultations were held to provide additional insights. Historically, Iraq's construction sector has disregarded risk management and restrictions of finance. The project hazards were assessed using surveys, expert comments, and exploratory research. Logistic challenges were also considered in the assessment process. Microsoft Excel made performance evaluations easier, and MAT F5 obtained the highest rating for sustainable materials in WPM's risk assessment. Furthermore, equipment output has emerged as a critical aspect in guaranteeing technical compliance. This study introduces a structured risk evaluation approach for improving sustainable construction practices in developing countries
Development and in vivo evaluation of a novel semi-solid nanostructured lipid carrier of fluticasone propionate for the treatment of atopic dermatitis using a quality by design approach
No full textFunding agency : Turkiye Bilimsel ve Teknolojik Arastirma Kurumu (TUBITAK).
Grant number : 117S967.Atopic Dermatitis (AD) is a chronic inflammatory skin condition that significantly affects patients' quality of life. This study focuses on the development of an innovative semi-solid nanostructured lipid carrier (NLC) dispersion containing fluticasone propionate (FP), aimed at enhancing therapeutic efficacy in the treatment of AD while minimising the systemic side effects associated with corticosteroids. The semi-solid NLC dispersions were prepared using a novel single-step preparation method. This method allows the formulations to maintain a colloidal particle size despite their high lipid content and semi-solid consistency. Particle size and drug release rate were identified as critical attributes, and Quality by Design (QbD)-assisted optimisation was carried out using computer-based modelling. The optimum formulation had an average particle size of 187.6 +/- 4.613 nm, which was within the targeted range. The polydispersitiy index (PDI) value of 0.229 +/- 0.019 indicates a relatively narrow size distribution. Furthermore, the amount of FP released from the optimum formulation at 24 h was 11.26 +/- 0.14 %, the highest among all the semi-solid NLCs prepared in the study. Results from the skin bleaching assay, paw oedema test, transepidermal water loss (TEWL) measurement, and histopathological evaluation in rats with induced chronic atopic dermatitis demonstrated both the enhanced therapeutic potential, and the favourable safety profile of the optimum formulation compared to conventional formulations. These findings suggest that semi-solid NLCs may serve as a promising alternative for the effective topical treatment of atopic dermatitis and other skin diseases