Project Description

Biological systems, including infectious diseases, are inherently complex and multi-scale in nature. Host-pathogen interactions span the molecular, organ and organism levels and occur on the order of seconds to years. Disease spread occurs through a series of immediate, local and global connections on the order of days to years. Effective control of infectious diseases requires a deep understanding of the complex interactions between these wide-ranging spatial and temporal scales. Experimental analysis across such vast scales is not feasible. On the other hand, multi-scale computational approaches provide ideal tools to integrate datasets across these scales to provide deeper insights and targeted biomedical interventions. To date, bridging the patient to population scales has been challenging in part due to the challenges associated with simultaneous simulation of multiple complex models at multiple scales. Thus, simplifying assumptions are commonly made, which has great danger to lead to significantly suboptimal or sometimes harmful disease intervention and control decisions. In this project we propose to integrate multi-scale modeling with statistical machine learning approaches to efficiently and rigorously extract key and interpretable features from smaller-scale simulations and adaptively transfer this information to larger-scale simulations. To better harness such efficient data exchange between scales, we also propose to develop active learning methods for efficiently characterizing spatiotemporal dynamics of complex systems based on limited experimental and real-world observations. Through the above two objectives, we will greatly expand our predictive analytics capability for biomedical systems science and engineering.

Start Date

April 2021

Postdoc Qualifications

Previous experience in mathematical modeling, machine learning, statistical methods.

Co-Advisors

Elsje Pienaar, Weldon School of Biomedical Engineering, epienaar@purdue.edu, engineering.purdue.edu/PienaarLab

Vinayak Rao, Department of Statistics, varao@purdue.edu, varao.github.io/ 

References

Pienaar, E., et al. :Comparing efficacies of moxifloxacin, levofloxacin and gatifloxacin in tuberculosis granulomas using a multi-scale systems pharmacology approach” PLoS computational biology (2019)

Tang, B., et al. Group-Representative Functional Network Estimation from Multi-Subject fMRI Data via MRF-based Image Segmentation Computer Methods and Programs in Biomedicine. Computer Methods and Programs in Biomedicine. (2019)