http://suspace.su.edu.bd/handle/123456789/4 2026-04-18T12:37:53Z 2026-04-18T12:37:53Z Hossain, Mohammad Sihad http://suspace.su.edu.bd/handle/123456789/2621 2026-03-31T04:13:18Z 2025-01-12T00:00:00Z

Design and Implementation of Inventory System for Local Shop Hossain, Mohammad Sihad Inventory management is a crucial aspect of business operations, particularly for small and local shops where efficient stock control directly affects profitability and customer satisfaction. Traditional manual inventory systems are often time-consuming, error-prone, and inefficient, making it difficult for shop owners to maintain accurate stock records and generate useful reports. This project presents the design and implementation of a web-based inventory management system for a local shop. The system is developed using ASP.NET Core MVC, C#, and Microsoft SQL Server to automate inventory operations such as product management, purchase tracking, sales management, and inventory reporting. The system ensures real-time stock updates by automatically adjusting inventory levels during purchase and sales transactions. The proposed system provides a user-friendly interface that allows authorized users to manage inventory data efficiently without requiring advanced technical knowledge. Testing results demonstrate that the system improves accuracy, reduces human error, and enhances overall operational efficiency. The system is suitable for small and local shops and can be extended in the future with additional features such as barcode integration and mobile application support.

2025-01-12T00:00:00Z Md., Al Mamunuzzaman Hredoy http://suspace.su.edu.bd/handle/123456789/2620 2026-03-31T04:07:07Z 2025-01-12T00:00:00Z

LungSightNet: Deep Learning Driven Lung Cancer Prediction Using Compact Convolutional Transformers Md., Al Mamunuzzaman Hredoy Accurate and interpretable medical image classification remains a critical challenge in computer-aided diagnosis, particularly under limited dataset conditions. Deep learning models often struggle to capture both local and global patterns simultaneously, and their black-box nature limits clinical trust. Incorporating attention mechanisms and explain- ability techniques can enhance both performance and interpretability. This research proposes a modified Self-Attention Compact Convolutional Transformer (SA-CCT) architecture designed to improve feature extraction and classification performance for lung cancer detection. The model integrates enhanced self-attention mechanisms and custom MLP blocks within the transformer framework, coupled with an improved patch based tokenization strategy that captures richer local and global features from grayscale CT images. To ensure interpretability, Grad-CAM explainability and segmentation based visualization modules are incorporated that enables spatial localization of discriminative regions and validation of model focus. The proposed SA-CCT model is trained and evaluated on a comprehensive lung cancer dataset that achieves 99% of classification accuracy, along with robust performance across per-class metrics and confusion analysis. These results show that the modified SA-CCT architecture is a very successful and easy to understand way to automatically diagnose lung cancer. It also provides a framework that may be used again and again for future research in medical image analysis and transformer-based categorization. Keywords: Lung Cancer Classification, CT Image, SA-CCT, CNN-Transformer Hybrid, Self-Attention, Explainable AI, Grad-CAM, Convolutional Tokenization, Custom MLP Block, Sequence Pooling.

2025-01-12T00:00:00Z Taspia, Momotaz http://suspace.su.edu.bd/handle/123456789/2619 2026-03-31T04:02:29Z 2025-01-12T00:00:00Z

Early Detection of Depression and Anxiety in Young Adults Using Machine Learning on Social Media and Survey Data Taspia, Momotaz Anxiety and depression among young adults have become serious public health concerns, yet early detection remains challenging due to the limitations of traditional mental health assessment methods. Conventional approaches rely mainly on self-reported questionnaires and face-to-face clinical evaluations, which are often conducted only after symptoms become severe. At the same time, young adults increasingly express their emotions, stress, and psychological distress through social media platforms such as Facebook, providing valuable real-world behavioral signals. However, most existing detection systems depend on a single data source, either social media text or psychological survey data, resulting in limited accuracy and an incomplete understanding of mental health conditions. Previous studies have showed that machine learning techniques can successfully identify linguistic patterns related to anxiety and depression from social media content. Similarly, standardized psychological assessment tools such as the Patient Health Questionnaire (PHQ-9) and the Generalized Anxiety Disorder scale (GAD-7) are widely recognized for measuring symptom severity through numerical scoring. Despite these advances, most prior research analyzes these data sources independently, with limited attention to integrated approaches that combine unstructured textual data with structured clinical information. To address this gap, This research establishes a highly effective hybrid framework for the early detection of anxiety and depression in young adults by integrating evidence from two complementary data sources. This study we proposeds a hybrid machine learning framework that integrates 6,877, Facebook posts with 103 there are 28 females and 75 male inside it self-report questionnaires based on the PHQ-9 and GAD-7 scales. A dual-model strategy was employed to accommodate the distinct characteristics of linguistic and numerical data. Different machine learning algorithms were evaluated for each data modality to identify optimal models. Experimental results demonstrate that Random Forest Machine (RF) achieves the highest accuracy 0.937821 for classifying anxiety and depression from Facebook text, while (SVM) provides accuracy 0.9905 superior performance for survey-based predictions. The fusion of these best-performing models significantly enhances overall detection accuracy, confirming that combining social media behavior with clinically validated psychological assessments offers a more reliable and comprehensive approach to early mental health risk detection among young adults. Keywords: Anxiety, Depression, Machine Learning, Social Media Analysis, Facebook posts, Hybrid model, Psychological questionnaires ,PHQ-9, GAD-7, Young Adults, Mental health assessment.

2025-01-12T00:00:00Z Jahangir, Hussen http://suspace.su.edu.bd/handle/123456789/2618 2026-03-31T04:00:45Z 2025-01-12T00:00:00Z

A Reliable and Efficient Approach to Suicidal Ideation Detection in a Low Resource Language Jahangir, Hussen Suicide is an endemic and disastrous global public health issue, necessitating the creation of scalable and forward-looking early detection methods beyond conventional clinical frameworks. Despite remarkable computational progress in high-resource languages such as English, the vast Bangla (Bengali) speaker population, ranging between 250 and 290 million worldwide, is underrepresented severely due to an existing computational imbalance characterized by data scarcity, inadequate linguistic content, and inherent problems such as affluent morphological richness, which hinders standard Natural Language Processing (NLP) methods. This research fills this technology gap by developing, evaluating, and rigorously validating a highly accurate, effective, and operationally robust Bangla Suicide Risk Classification system from user-generated digital text with real-world applicability in low resource healthcare environments. Empirically confirming its assertions through an elite, clinically annotated corpus, this research demonstrates that Character Ngram TF-IDF Vectorization is the optimal feature engineering method, outperforming word-level embeddings by being more adept at dealing with data sparsity. Massive benchmarking across thirteen disparate Machine Learning (ML) and Deep Learning (DL) models obviates the critical Deployment Paradox, signifying a trade-off between predictive performance and computational cost. The best safety performance (Recall: 0.9280, 92 False Negatives) was achieved by the Bi-directional Long Short-Term Memory (BiLSTM) model but at the expense of crippling latency (5.23 seconds), rendering it useless for real-time triage. On the other hand, the light-weight RidgeClassifier (RC) with the same feature representation obtained an equivalent Recall of 0.9170 (106 False Negatives) with near zero latency (0.001 seconds), which is the Optimal Deployable Triage System for large-scale real-time intervention. This paper highlights that interpretable and computationally efficient ML models can outperform state-of-the-art DL architectures in real-world deployment scenarios. Besides, it encourages ethical deployment with interpretable feature weights and Dynamic Threshold Tuning (Human-in-the-Loop) for system sensitivity tuning to adapt to changes in resources in an effort to ensure a sustainable, safe, and effective suicide prevention tool for the Bangla-speaking populations of the world.

2025-01-12T00:00:00Z