🛡️ TrustAlert: News Time Series Anomaly Detection
Detecting anomalies in disease-related news coverage using advanced time series analysis
This tool analyzes temporal patterns in disease-related news coverage to identify potential outbreaks or unusual events. By detecting anomalies in the frequency of disease mentions, we can help public health officials spot emerging health concerns early.
Welcome! 👋
This demo is part of the TrustAlert: Empowering Public Health with Real-Time Insights and Future Preparedness project, where we detect unusual patterns in disease-related news coverage that might indicate potential outbreaks or emerging health concerns.
AI for early health alerts, powered by time series analysis and anomaly detection.
🔍 What This Demo Does
- 📊 Time Series Visualization
Upload your CSV file containing dates and disease mention counts, and visualize the temporal patterns using interactive Plotly charts. - 🔍 Anomaly Detection
Choose from multiple detection methods to identify unusual patterns in your time series:- LSTM: Uses deep learning to model sequential data and detect anomalies based on deviations from predicted patterns
- ARIMA: Employs statistical methods to forecast time series and identify anomalies by comparing actual values to predictions
- IQR: Flags anomalies by identifying data points that fall outside the interquartile range
🧪 Try It Yourself
- 📈 Upload a CSV file with two columns: dates in the first column and disease mention counts in the second
- 🎯 Click "Plot Time Series" to visualize your data
- 🔍 Select an anomaly detection method from the dropdown
- ⚙️ Configure the detection parameters:
- For LSTM method:
- k: Controls sensitivity (1-3). Higher values mean stricter anomaly detection.
- Percentile: Threshold percentile for anomaly detection (0-100).
- Threshold Method: Choose how to calculate anomaly thresholds:
- IQR-based methods: Compare predictions with actual values using different metrics
- Percentile-based methods: Use statistical thresholds on prediction errors
- For ARIMA method:
- k: Sensitivity multiplier for standard deviation-based thresholds (1-3).
- For IQR method:
- k: IQR multiplier (1-3). Higher values detect more extreme outliers.
- For LSTM method:
- ⚡ Click "Detect Anomalies" to identify unusual patterns in your time series
📋 Example Dataset
Try out the tool with our sample dataset:
- Dataset:
mpox.csv- News coverage time series for Monkeypox/Mpox outbreak - Time Period: Daily counts from early 2022
- Recommended Settings:
- Method: LSTM
- k: 1.5
- Percentile: 95
- Threshold Method: "IQR on |ground truth - forecast|"
- Expected Results: The analysis should identify significant spikes in news coverage that corresponded to major outbreak events and public health announcements during the 2022 Mpox outbreak.
This tool combines time series analysis and anomaly detection to help identify potential disease outbreaks based on news coverage patterns. The results can be used to alert public health officials about emerging health concerns. 💡