Faculty Scholars

About

Shyam Ranganathan is an applied statistician in the School of Mathematical and Statistical Sciences. He has a Ph.D. in Mathematics with special focus on Applied Mathematics and Statistics, and has worked in academia over the last 6 years, five of them at Virginia Tech as assistant professor. He primarily works on statistical modeling of high-dimensional, time-series and spatio-temporal datasets and my application areas span engineering, health sciences, sociology, economics, sustainability science, public policy etc. Specifically, within healthcare-related research, he was co-PI of an NIH proposal that looked at the effects of surface mining on maternal outcomes with epidemiologists, geographers and public health researchers. This work has resulted in the publication of multiple articles in reputed journals on both the modeling and epidemiological aspects of the problem. Another research project that he was involved in was the study of self-injurious behavior of autistic children. This was originally funded by the NSF, though the later analysis that Ranganthan conducted was funded by a local University grant. The research team showed how predictive modeling of this behavior can help address longstanding issues with the care of these children, and this work was published in Nature Scientific Reports. In recent months, he has been working with doctors at PRISMA Health on two projects – one for a study on the effects of VR interventions in palliative care for cancer patients, and the other on using AI-based methods for remote patient monitoring of heart patients. Both projects were initiated by collaborators at Industrial Engineering (Dr. Kapil Madathil and Dr. Sudeep Hegde) and Architecture (Dr. Anjali Joseph) and are ongoing.

View Ranganathan's Faculty Profile.

How their research is transforming health care

Statistics, and especially core topics such as design and analysis of experiments, the use of time-series methods etc. are pervasive in healthcare research. Machine learning and artificial intelligence is also gaining ground as the use of large, complex datasets collected using multiple sensors is becoming ubiquitous. In these complex problems, it is tempting to use “black boxes” that potentially could predict outcomes in any given problem. However, the misgivings following the inability of the supercomputer Watson to “solve cancer” as it was advertised a decade ago, has created a more mature understanding of how these tools can be applied to aid the healthcare professional rather than to replace them. Ranganathan’s research focuses on using inputs both from the multitude of sensors that technology has created for the healthcare system, and the expert who has trained for years to understand a specific problem to create a synthesis that will utilize the best of both worlds to address key healthcare problems. He believes this kind of “human-in-the-loop” machine learning is a key paradigm in coming years and will involve more serious engagement between statisticians, computer scientists, data scientists, and, most importantly healthcare professionals. In one of his projects with PRISMA Health, the team initiated this idea by sitting not just with the doctors but also nurses and caregivers to understand complexities in remote patient monitoring – non-compliance with protocol, non-uniform application of rules etc. are as much of a confounder in remote monitoring models as are other technical or data outliers.

Health Research Expertise Keywords

Faculty Scholar, Statistical modeling, time-series, spatio-temporal data, multi-sensor fusion, machine learning, AI in healthcare, Bayesian modeling, causal analysis