Thesis Report........docx
| dc.contributor.author | Taspia Tazri Chaity | |
| dc.date.accessioned | 2025-11-23T03:51:56Z | |
| dc.date.available | 2025-11-23T03:51:56Z | |
| dc.date.issued | 2025-09-19 | |
| dc.description.abstract | Dengue fever is a viral infectious disease caused by four distinct serotypes, namely DENV 1, DENV 2, DENV 3, and DENV 4, and it spreads primarily through the bite of an Aedes mosquito that has previously fed on an infected person. It has caused great problems in diagnosis and treatment of the disease at the early stages of it since it has spread to almost all parts of the world thanks to over-urbanization global warming and poor measures to counter the vectors leading to a mixed picture of Mild Dengue Fever (DF) and severe forms such as Dengue Hemorrhagic Fever (DHF) and Dengue Shock Syndrome (DSS). Disease severity classification is by far the most significant tool to minimize complications and maximize care provided to the patient, in addition to efficient use of healthcare resources. 70 14 This study is grounded in 1,157 records from Cumilla, Noakhali, and Dhaka in Bangladesh, incorporating clinical and physiological details, demographic information, contextual factors, and cases representing both dengue and non-dengue conditions to ensure balanced representation. Using machine learning and deep learning techniques, the records were classified into four categories: Mild Dengue Fever, Dengue Hemorrhagic Fever, Dengue Shock Syndrome, and No Dengue. To enhance predictive accuracy, a novel bybrid RF-LSTM model was developed by integrating the Random Forest algorithm with the Long Short-Term Memory network. The Random Forest approach was employed for effective feature extraction, while the LSTM network captured and modeled the complex, non-linear temporal patterns in the data. Experimental findings demonstrated that the hybrid RF-LSTM model achieved superior performance, reaching an accuracy of 96.26%, and it consistently outperformed conventional models in terms of accuracy, precision, and recall. Furthermore, the generalizability of the model was validated using unseen datasets, confirming its robustness and reliability for real-world applications. Beyond individual-level classification, the study incorporated geographic analysis to generate risk zone maps, which assist in identifying high-risk areas and formulating targeted intervention strategies. These maps provide valuable insights for public health planning, efficient allocation of resources, and proactive management of dengue outbreaks. Overall, the study establishes a comprehensive, information-driven framework for diagnosis, severity classification, and geographical risk assessment of dengue, while demonstrating how computational modeling combined with epidemiological expertise can enable timely disease detection. Moreover, the approach strengthens population health surveillance and offers a scalable solution adaptable to other vector-borne diseases, thereby contributing to the development of sustainable, data-driven healthcare systems. Keywords: Dengue, Outbreak, Risk zones, Severity, Public health, Public awareness. | |
| dc.identifier.uri | https://dspace.nstu.ac.bd/handle/123456789/107 | |
| dc.title | Thesis Report........docx |