Machine Learning/AI – Insight Net

Machine Learning/AI

SISMID 2026: Human – AI Teaming Approach to Infectious Disease Modeling
This module introduces key concepts in infectious disease modeling alongside core principles of human–AI teaming. Students will begin with a guided walkthrough of AI prompting using an example dataset, demonstrating how to build a full forecasting pipeline step-by-step. This demo is followed by hands-on labs where students apply the same prompting strategies to a more…
Trainings Market Analysis
Systematic analysis on the market to analyze the landscape, demand, and need for different types of trainings.
Simulation Games
Develop serious simulation games and planning to host workshops focused on these games.
Microcertifications
A system of online training modules and microcertifications
Embedding postdocs and tool builders
Embedding postdocs and tools building from UCSD
Multitask Model to Forecast Patient’s Next Unit and Remaining Length of Stay
Length of Stay prediction and unit transfer prediction tool. Scenarios(s) Modeled: In this multi-task learning model, we train the model to predict:Task-1: patient’s unit label in next time step. Task-2: patient’s remaining length of stay.The model inputs get updated on a daily basis and generates prediction accordingly.
Attention-Based Models for Snow-Water Equivalent Prediction
Transformer architectures for spatio-temporal prediction (or synthetic data generation/imputation). Scenarios(s) Modeled: Predicting the SWE value for multiple SNOTEL locations in the Western US using the Attention Models
Flusion model code
Ensemble of statistical (Bayesian autoregressive) and machine learning (gradient boosted quantile regression) time series models, using a combination of influenza ground truth data inputs (NHSN, FluSurv-NET, ILINet) to calibrate model parameters. The inclusion of the latter two data sources is meant to augment the first, as NHSN only has a limited amount of historical data…
Deep Learning Model for Epidemic Forecasting (DLEF)
Utilizies electronic health records, social media data, and community level data to predict current infectious and forecast epidemic curves
Machine Learning and Probabilistic Approaches for Forecasting COVID-19 Transmission and Cases
The purpose of this project is to employ machine learning and probabilistic approaches for real-time forecasting of infectious disease transmission (COVID-19) and positive test counts at the local level using electronic health record (EHR) data from Prisma Health and MUSC.