Retrieval-augmented generation (RAG) supports policy question answering by grounding responses in institutional documents. This study describes a campus assistant developed with Azure OpenAI GPT-4 and an Azure Cosmos DB knowledge store containing university regulations. The evaluation covers 51 regulation documents organised into five categories and a question-answer set comprising 1,742 items. Human reviewers classified 1,644 system replies as correct, yielding an overall response accuracy of 94.37%. Response accuracy is an aggregate human-judged correctness measure, not exact match. Because the study did not retain item-level evaluation materials, it does not report separate measures of retrieval quality, citation fidelity, unsupported content, latency, or user experience.
Acute Myeloid Leukaemia (AML) is an aggressive form of leukaemia with several subtypes. Timely and accurate detection of these subtypes is needed for better treatment outcomes. Artificial intelligence helps in diagnostics, but morphological variability, inconsistent imaging conditions, and dataset imbalance hinder performance. Existing methods depend mostly on handcrafted features and traditional classifiers with limited subtype coverage. This study proposes TrustNet, a deep learning framework built on a dynamic channel-spatial attention network (DCSA-CNN) for classifying morphology-based AML subtypes. The DCSA backbone integrates adaptive channel scaling, lightweight spatial attention, and residual attention fusion to enhance feature learning over fixed attention mechanisms. Comparative evaluation validates its effectiveness across baseline CNNs, ablation variants, attention-based architectures, and pre-trained models. TrustNet extends this framework by integrating test-time augmentation and a trust-aware strategy that filters low-confidence predictions. Results show strong performance on retained samples, demonstrating improved reliability of the proposed framework.
