Avian Flu and Zoonoses Overview

Lecture 8: Integrated Data Analysis and Outbreak Prediction: Avian Flu

• The lecture is about integrated data analysis and outbreak prediction, focusing on avian flu.
• The presenter is Marina Treskova, PhD, Head of Research Group, Eco-Epidemiology.
• The research team is associated with the Heidelberg Institute of Global Health & Interdisciplinary Centre for Scientific Computing, Heidelberg University.
• Website: hei-planet.com

Spillover and Zoonotic Disease Emergence

• Spillover is a key step in zoonotic disease emergence.
• Spillover cascade: The process starts with infection of wildlife, then shedding of the pathogen (animal-to-animal transmission), spillover to humans or other animals, followed by spread within human populations.
• Various pathways for zoonotic spillover:
  • Direct transmission
  • Vertebrate intermediate host
  • Invertebrate intermediate host
  • Environment

Known Zoonotic Diseases

• The number of included diseases is 203.
• Of these, 70 are viral, 47 bacterial, and 74 parasitic.
• The total number of outbreak events is 13985.
• Zoonotic spillover is a key public health challenge

WHO Priority Diseases

• Diseases highlighted by WHO include Ebola virus, Lassa fever, Crimean-Congo haemorrhagic fever, Middle East respiratory syndrome coronavirus (MERS-CoV), Nipah virus infection, Rift Valley fever, and Zika virus.

Clusters of Zoonotic Spillover

• Various factors are associated with zoonotic spillover.
• Factors include viral adaptation, climate and vectors, food & livestock, environmental contact, global movement, socioeconomics, and land conversion.

Climate Change and Human Pathogenic Diseases

• Over half of known human pathogenic diseases can be aggravated by climate change.
• Key factors associated with climate change are warming, precipitation, floods, droughts, storms, and natural cover change.

Land Use and Ecosystem Degradation

• Ecosystem degradation and human encroachment contribute to zoonotic spillover risk.
• Ecosystem restoration and biodiversity conservation are crucial to reduce spillover risk.

Biodiversity Loss and Dilution Effect

• Lyme disease is used as an example
• High biodiversity reduces reservoir density. This makes zoonotic disease transmission less likely.
• Low biodiversity increases reservoir density making zoonotic disease transmission more likely.

Social Drivers

• Globalization and environment: Deforestation, mining and dams, and ecosystem degradation can change vector and non-human host habitats. International travel and trade, urbanization, and population displacement spread pathogens and increase risk factors.
• Sociodemographic factors: Children, the elderly, and pregnant women are more vulnerable. Quality of housing, exposure to vectors, baseline incidence of disease, population health status, and humanitarian crises impact vulnerability.

Public Health Systems

• Surveillance, early warning systems and vector control are important to prevent disease.
• Access to and quality of healthcare can affect severity of infections.
• Research is key to effectively control and prevent these diseases.

One Health Approach to Zoonoses and Prevention

• This approach emphasizes reducing disease risk at the source, not just responding to outbreaks after they have happened.

Avian Flu Model

• The avian flu model includes several features, and focuses on how well these can predict future outbreaks.
• 3 models were built for this study.
• Model 1: focuses on eco-climatic and socio-economic variables, and determining predictability levels.
• Model 2: focuses on wild bird outbreaks.
• Model 3: focuses on wild bird outbreaks but removes bird labels

Model Metrics

• Evaluation metrics for different models. Logloss was used to measure the effectiveness of models.

Main Results

• Model 1, 2, and 3 results are highly accurate when predicting Avian Influenza outbreaks in Europe.
• The model is climate sensitive.
• The key climate indicators are the temperature of the coldest month, the mean temperature of quarter two, and the minimum temperature.
• Climate, Environmental, and Bioclimatic features are most important for the model.
• Trade and Economic factors are also significant.

SHAP vs Importance

• Differences between SHAP feature importance and XGBoost standard feature importance
• SHAP considers feature correlations
• XGBoost does not
• SHAP provides both local and global feature importance. XGBoost's feature importance is mainly global.

Test your knowledge on Avian Flu, its classification, and the environmental and socioeconomic factors that influence its outbreaks. This quiz also covers the One Health approach and terminology related to zoonotic diseases. Ideal for students and professionals interested in infectious diseases.