AI4AI uses machine learning to classify avian influenza host species from DNA sequences.
problem Classifying avian influenza host species from DNA sequences to reduce emergency response time.
method Quantitative methods using machine learning and deep learning.
result Best deep learning models achieve top-1 classification accuracy of 47%, and top-3 classification accuracy of 82%.
Study of influenza A virus spread using mathematical equations.
problem Understanding the spread of influenza A virus infection.
method Mathematical model and analysis of dynamical system.
result Surface trajectories and asymptotic behavior of the system.
Modeling influenza spread using feature engineering and international flow deconvolution.
problem Predicting and mitigating influenza spread through feature extraction and international flow analysis.
method Discrete Fourier Transform, matrix completion, SVM, autoencoders, PCA, deconvolution of international flow.
result Significant environmental and economic features are crucial to influenza mortality.
Algorithm predicts zoonotic virus emergence with high accuracy.
problem Precise spatio-temporal prediction of zoonotic virus emergence.
method Machine inference using protein sequence databases.
result Quantitative indicators of jump risk from genotypic changes.
Adaptive ensemble improves flu forecasts with minimal data.
problem Accurate flu forecasts to help public health.
method Adaptive stacking of ensembles, changing model weights weekly.
result Adaptive ensemble outperforms static ensembles in flu forecasts.
GRUs predict flu at state and city levels with lower error.
problem Influenza prediction at multiple spatial resolutions.
method Gated Recurrent Unit (GRU) neural network, real-time search data.
result GRU outperforms state-of-the-art methods for flu prediction.
Accurate real-time tracking of influenza outbreaks helps public health officials make timely and meaningful decisions that could save lives. We propose an influenza tracking model, ARGO (AutoRegression with GOogle search data), that uses publicly available online search data. In addition to having a rigorous statistica…
Flusion combines multiple data sources to improve flu forecasts.
problem Accurate flu predictions to improve public health actions.
method Ensemble model combining gradient boosting quantile regression and Bayesian autoregressive models.
result Flusion was the top-performing model in the CDC's influenza prediction challenge.
Emergenet predicts animal influenza strain emergence, outperforming current methods.
problem Limited ability to quantitatively assess animal influenza strain emergence.
method Infer digital twin of sequence evolution using 220,151 HA sequences.
result Emergenet predictions outperform WHO seasonal vaccine recommendations and CDC IRAT scores.
Accurate real-time monitoring systems of influenza outbreaks help public health officials make informed decisions that may help save lives. We show that information extracted from cloud-based electronic health records databases, in combination with machine learning techniques and historical epidemiological information,…
New method uses Transformers for flu forecasting.
problem Forecasting influenza-like illness trends.
method Transformer-based machine learning models with self-attention.
result Forecasting results are competitive with state-of-the-art methods.
MutaGAN predicts mutations of evolving protein populations using GANs.
problem Predicting mutations in evolving protein populations.
method Generative adversarial networks (GANs) with recurrent neural networks (RNNs).
result MutaGAN generates complete protein sequences with mutations.
Bayesian consensus improves accuracy of forecasts from miscalibrated sources.
problem Aggregating predictions from miscalibrated and noisy sources.
method Bayesian approach to adjust for bias and noise, using hierarchical models.
result Bayesian consensus estimator is unbiased and more efficient than alternatives.
Improves flu prediction by blending environment and population info.
problem Challenges in using data from one environment in another due to feature variability and population subgroup differences.
method Population-aware hierarchical Bayesian domain adaptation framework with multiple invariant components.
result Model improves flu prediction in new environments with unlabelled data.
Paper proposes a graph neural network for accurate long-term ILI prediction.
problem Limited long-term prediction performance and spatio-temporal dependency in existing models.
method Cross-location attention based graph neural network (Cola-GNN) for time series embeddings and location aware attentions.
result Proposed method shows strong predictive performance and interpretable results for long-term epidemic predictions.
Study uses Instagram data to predict flu-like illnesses.
problem Influenza surveillance challenges due to resource limitations.
method Machine learning models trained on Instagram data, including visual content.
result Best nowcasting model had an MAE of 11.33 and correlation of 0.963.
Accurate and reliable predictions of infectious disease dynamics can be valuable to public health organizations that plan interventions to decrease or prevent disease transmission. A great variety of models have been developed for this task, using different model structures, covariates, and targets for prediction. Expe…
Estimates effect sizes and power from a pilot experiment.
problem Estimating the distribution of effect sizes in multiple testing settings.
method Uses an inexpensive pilot experiment to estimate effect sizes and power.
result Simple and computationally efficient estimator guarantees the number of discoveries.
WeatherFormer learns robust weather features from small datasets.
problem Modeling complex weather dynamics from limited data.
method Pretrained transformer encoder on large satellite dataset, with spatiotemporal encoding.
result State-of-the-art performance in county-level soybean yield prediction and influenza forecasting.
This paper uses social signals to improve cryptocurrency price forecasting.
problem Improving cryptocurrency price forecasting using social signals.
method LSTMs trained on historical price data and social data from GitHub and Reddit.
result Social signals reduce error in forecasting cryptocurrency prices, especially for Bitcoin.
Improved county-level COVID-19 forecasting model using LSTM and data augmentation.
problem Accurately forecasting county-level COVID-19 cases to optimize medical resources.
method Adapted TDEFSI-LONLY model, utilized LSTM, data augmentation, and inter-county mixing.
result CLEIR-Net model provides better forecasts than TDEFSI-LONLY.
Social media services such as Twitter are a valuable source of information for decision support systems. Many studies have shown that this also holds for the medical domain, where Twitter is considered a viable tool for public health officials to sift through relevant information for the early detection, management, an…
A new method combines ANN and Laplace for fast Bayesian inference in ODE models.
problem Bayesian inference for ODE systems with non-analytical solutions is computationally expensive.
method Hybrid approach using ANN for tractable likelihood and Laplace approximation.
result Effective posterior inference with improved computational cost compared to traditional methods.
Deep learning models outperform traditional methods in automated chief complaint classification for syndromic surveillance.
problem Improving accuracy and speed of automated classification of emergency department records for outbreak detection.
method Implemented two LSTM and GRU models compared to MNB and SVM classifiers trained on 3.6 million de-identified records.
result RNN models outperform bag-of-words classifiers, especially for chief complaints.