تقدم هذه الورقة بحثاً حول إطار عمل يعتمد على التعلم العميق للتصنيف متعدد الملصقات لأمراض الصدر باستخدام صور الأشعة السينية وبنية EfficientNet. يعالج هذا الإطار تحدي تشخيص الحالات المتعددة التي تتشابه في المظهر وغالباً ما تحدث معاً. يستخدم إطار العمل تقنية التعلم بنقل الخبرة (Transfer Learning)، مع إجراء ضبط دقيق لعدة متغيرات من نموذج EfficientNet (من B0 إلى B4) باستخدام مجموعة بيانات NIH ChestX-ray14. كما يوظف البحث تقنيات Focal Loss و Stochastic Weight Averaging لمعالجة اختلال توازن البيانات وتحسين قدرة النموذج على التعميم. يعتمد هذا النهج على آليات الانتباه (Attention Mechanisms) وتقنية Grad-CAM لجعل النتائج أكثر قابلية للتفسير، مما يثبت أن التوقعات تستند إلى ميزات سريرية هامة. حقق النموذج درجات عالية في مقياس ROC-AUC للحالات الأكثر وضوحاً مثل الانصباب (0.8545)، وتضخم القلب (0.8003)، والاسترواح الصدري (0.7820). ومع ذلك، واجه النموذج صعوبة أكبر في تصنيف الحالات الأكثر دقة مثل الارتشاح، والعقيدات، والالتهاب الرئوي. يؤدي إطار العمل بشكل يضاهي أو يتفوق على العديد من نماذج الشبكات العصبية التلافيفية (CNN) القائمة من حيث الدقة والكفاءة. يوفر هذا البحث سير عمل قوياً وقابلاً للتكرار لتطبيقات الأشعة، مع إمكانية استخدامه في مهام تشخيصية أخرى. كما تشير الدراسة إلى بعض المعوقات، مثل اختلال توازن الفئات والحاجة إلى دعم أفضل لاتخاذ القرارات السريرية، وتقترح تحسينات مستقبلية مثل تقنيات أفضل لتعزيز البيانات ونمذجة عدم اليقين.

Abstract

One of the most challenging  skills for English as a Foreign Language (EFL) learners is often the ability to speak, especially in environments where opportunities for authentic English exposure are scarce. This study focuses on prevalent linguistic errors found in the spoken English of third semester EFL students at the Faculty of Arts, Gharian University. . The three primary types of errors that are examined are phonological, grammatical, and lexical. The study also explores  the possible reasons behind these errors, such as first-language (L1) interference, limited English exposure, inadequate speaking practice in the classroom, fossilization, and emotional factors. Students often face difficulties with English phonemes that do not exist in Arabic, as well as with the correct grammatical forms, and suitable vocabulary choices, according to the results. The study emphasizes how important it is to implement focused interventions to improve students' spoken accuracy and fluency. It suggests pedagogical recommendations to enhance the accuracy and fluency of students’ spoken English, including increasing the number of opportunities for communicative practice, targeted pronunciation training, and the incorporation of authentic listening resources.

Keywords, English as a Foreign Language, linguistic errors, speaking skill

Abstract

The proper management of biological waste in laboratory settings is a critical aspect of biosafety in microbiological and biomedical laboratories. Beyond protecting lab workers, biological waste management safeguards public health, protects the environment, and ensures compliance with legal and regulatory frameworks. Segregation, decontamination of liquid waste, decontamination of solid waste, and techniques for inactivating toxins are all main aspects of biological waste management that will be outlined, aligned, and summarized with the main institutional biosafety protocols for biological waste management, published by major authorities, including CDC and WHO. This research found that the information from the two institutions complements each other. Therefore, this brief review aimed to compile and unify information from both institutions to construct a single, coherent text that provides a practical summary for medical laboratory professionals seeking to enhance the implementation of international, harmonized protocols that strengthen biosafety, environmental stewardship, and regulatory compliance

This study focused on documenting flowering plant species in Wadi Al-Yaquib in Gharyan City . The plants were collected between March and May 2024 , dtied, and transferred to the herbarium for classification, During field visits,42 plant species were collected and classified, belonging to 40 genera distributed across 21 families. Of these, and 34 genera were classified under 34 dicotyledonous species, represnting 85 % of the total, and distributed across 18 families, Also 6 genera were classified under speceis of monocotyledonous plants, These species reprsent a percentage of 15% distributed over a number of three plant families, The analysis of flora in relation to the total number of species and genera belonging to each family revealed that the family. Astraceae was the largest family in class dicotyledonus opsida As a comparsion between the current study and the previous studies, this study drew attention to a plant species potentially endangered due to environmental pressures and overgrazing.