Integrating Bayesian stats with machine learning for time-since-event prediction

The MLAI Meetup is a community for AI researchers and professionals which hosts monthly talks on exciting research. Our format is:

• 6:00 - 6:20: Socializing
• 6:20 - 6:40: Announcements and AI news
• 6:40 - 7:40: Talk(s) and Q&A
• 7:40 - 8:00 Networking
• 8:00: Head to the nearest pub for dinner

Joint event with the Statistical Society of Australia!

Felicia Bongiovanni: "Inferring Time of Infection Based on Antibody Data Through a Hierarchical Bayesian Framework"

Abstract: Time-since-event prediction is a powerful tool to reconstruct histories of an event. It is commonly used in biomedicine to predict the beginning of tumor growth, to predict how long ago someone was infected with a disease, and even to predict the time since death for criminal investigations. Inference through Bayesian statistics and mathematical modelling can allow for time-since-event prediction and provide insight into the mechanisms driving an event. Machine learning is also used for time-since-event prediction but lacks the mechanistic insight Bayesian inference can provide. However, Bayesian inference, especially through Markov Chain Monte Carlo methods, can be computationally expensive and time-consuming. In this talk, I describe a method for combining Bayesian inference with machine learning to improve computational efficiency while also maintaining mechanistic insight. This is demonstrated through the case study of estimating time-since-infection using antibody data.

Speaker bio: Felicia is a PhD candidate at The Walter and Eliza Hall Institute of Medical Research where she is conducting mathematical modelling of the immune response to understand more about infectious disease dynamics. In another life, she was a data analyst for the National Health Funding Body in Canberra through the Australia Bureau of Statistics Data Graduate Program.