Search
Loading

About me

I have been conducting studies on high-throughput cytometry, high-content imaging, biological image analysis, biological pattern recognition, and applications of statistical machine learning in cell biology, cancer research, and bioengineering. My research’s underlying, the long-term theme is the use of innovative data-science tools (and specifically machine-learning methods) to reduce complex phenotypic information into essential statistical models and gain insight into biological processes characterized by incomplete and noisy data. The major project that exemplifies the highly interdisciplinary nature of this inquiry and the inherently collaborative nature of my research is the ongoing development of a non-exhaustive learning framework (a set of methods addressing problems of learning with complex, noisy, and highly incomplete training data), in which he has been continuously involved over the last seven years.

The initial work initially inspired by a practical pathogen-detection problem (research sponsored by USDA-ARS) led to the development of a unique feature-extraction system capable of quantifying phenotypic changes in bacterial colonies. Subsequently, this research agenda expanded to address a broader issue of detection and/or classification of phenotypes in the absence of a complete model of the data source. The methodology created in collaboration with Dr. M Murat Dundar (IUPUI) in the first period of the study relied on a density-based approach, which used class-conditional likelihoods to detect new entities in samples. The success of this simple design encouraged further investigation, and the study was subsequently expanded to deliver a first universal paradigm for learning with incomplete or partially observed information via non-parametric Bayesian self-adjusting models. Clinical practitioners have recognized the enormous potential of this advanced strategy, and the technique has been recently successfully applied for automated diagnosis of acute myeloid leukemia (AML) and prediction of disease progression (a collaboration involving Dr. Paul Wallace).

The described study, seamlessly integrating over its course components from machine learning, statistics, microbiology, cell biology, and immunology, demonstrates the crucial role of data science in modern biomedical research and illustrates the importance of multidisciplinary approaches reaching beyond established disciplines and narrowly defined fields. I firmly believe that his ability to translate complex problems of quantifying biological responses into reducible mathematical constructs not only allows him to pursue an original and highly innovative research agenda but also enables me to serve as a key scientific intermediary and idea conduit between diverse groups of collaborators.

Publications and citations

See my publication list here.

Published Items in Each Year
Citations in Each Year

Page updated on 2022-01-17 13:38:20 -0500