Logo Lanfrica

AI-Powered Classification and Early Detection of Dengue Virus Lineages for Timely Public Health Response

Domaine:

healthcare

Type de record:

postermodel
Créateur:
NyaTegally, HouriiyahXav
Éditeur:
Zenodo
Hôte:avatar
Poster presented at the Deep Learning Indaba 2025 in Kigali, Rwanda. It summarises an alignment-free machine learning and deep learning pipeline that classifies Dengue virus (DENV) genomes into lineages and flags potential new variants. Genomes are encoded with k-mer counts, Frequency Chaos Game Representation (FCGR), and one-hot encoding, then classified with Random Forest, XGBoost, LightGBM, hierarchical classifiers, a 1D-CNN with self-attention, and a 2D-CNN, all reaching over 97% accuracy and/or F1-score. Class imbalance is handled with SMOTE. A Variational Autoencoder generates synthetic genomes, and a soft-voting ensemble flags candidate new variants rapidly, in agreement with the Genome Detective typing tool. 

Tags

Dengue Virusgenomic surveillancelineage classificationDeep Learningvariant detectionDeep Learning Indabaalignment-free methodsone-hot encodingself-attentionvariational autoencoder+5

Licenses

Creative Commons Attribution 4.0 Internationalhttps://creativecommons.org/licenses/by/4.0/legalcode

Similaires

AI-Powered Image Recognition for Early Detection of Pests in Medicinal Crops

Dalam beberapa tahun terakhir, perkembangan Kecerdasan Buatan (AI) sangat pesat di sektor pertanian,

Early genomic detection of cosmopolitan genotype of Dengue virus serotype 2, Angola, 2018

International audience We used portable genome sequencing to investigate reported den

Early genomic detection of the Cosmopolitan Genotype of Dengue Virus 2 in Angola, 2018

Abstract We used portable genome sequencing to investigate reported dengue virus t

Implementing epidemic intelligence in the WHO African region for early detection and response to acute public health events

Abstract Epidemic intelligence activities are undertaken by the WHO Regional Office for Africa

Dengue Outbreaks in Sudan: Epidemiology, Clinical Characteristics, Serotype Distribution, and Public Health Response.

This scoping review aims to systematically identify, map, and synthesise the breadth of primary rese

Adopting Zero-Trust Security Models in Nigeria's Critical Infrastructure: AI-Powered Detection and Response