Nov 23, 2025
BioCast AI
Mohammed Usman, Igor Muzin
BioCast AI is a deep learning powered system designed to forecast global disease outbreaks using neural sequence models and advanced machine learning ensembles. Our approach integrates three architectures RNN LSTM and Bidirectional LSTM to capture long term temporal patterns in more than forty years of global disease data. In parallel we developed a feature rich ensemble model using XGBoost LightGBM Random Forests and Gradient Boosting to enhance predictive accuracy and stability.
We trained models on cleaned and imputed WHO style datasets covering 1980 to 2024 generating country specific and disease specific predictions for 2025. BioCast AI also includes a full analytical pipeline with PCA correlation analysis, temporal visualization clustering and dashboards to support epidemiological decision making.
The system achieved strong performance across models with ensemble methods outperforming individual networks in R² and RMSE scores. Outbreak risk levels were derived using predicted case trajectories, trend ratios and statistical confidence intervals. BioCast AI demonstrates how deep learning and ML ensembles can support early warning systems, public health forecasting and global disease surveillance.
No reviews are available yet
Cite this work
@misc {
title={
(HckPrj) BioCast AI
},
author={
Mohammed Usman, Igor Muzin
},
date={
11/23/25
},
organization={Apart Research},
note={Research submission to the research sprint hosted by Apart.},
howpublished={https://apartresearch.com}
}
