Chandan K. Reddy - US grants

Hospitalizations account for more than 30% of the $2 trillion annual cost of healthcare in the United States. As many as 20% or more of all hospital admissions occur within 30 days of a previous discharge. Such rehospitalizations are not only expensive but are also potentially harmful, and most importantly, they are often preventable. Providing special care for a targeted group of patients who are at a high risk of rehospitalization can significantly improve the chances of avoiding rehospitalization. Estimating the predictive power of the clinical data collected during the hospitalization of a patient and effectively making predictions from such diverse patient records requires new analytical models. This project develops a 'rehospitalization analytics' framework that can identify, characterize and reduce the risks of rehospitalization for patients using a wide range of electronic health records. Specifically, the research objectives of this project are to develop: (i) integrated models that can effectively leverage multiple heterogeneous patient information sources and transfer the acquired knowledge about rehospitalization between different hospitals and patient groups in the presence of only few patient records, (ii) novel adaptable time-sensitive models that make predictions of the risk estimates in the presence of inherent concept drifts in the clinical data, and (iii) new regularization methods that can effectively extract the population-specific risk factors despite the presence of multiple correlations and grouped categorical clinical predictors. The methods are evaluated using heart failure patient records collected at the Henry Ford Health System in Detroit. The performance of the proposed models is compared against the state-of-the-art statistical and clinical tools that are currently applied for risk prediction.

This project aims to provide a comprehensive, accurate, and timely assessment of risk of rehospitalizations, and has the potential to direct more aggressive treatments towards specific high-risk patients. Predictive models developed in this project could be widely adopted and have nation-wide impact because the source data is often available at the hospitals. This has the potential to improve the lives of patients, by reducing exacerbations, and reducing overall health care costs by reducing the number of hospitalizations. The computational models developed in this project could also be applied to other chronic diseases that have high rates of utilization and could benefit from improved targeting of intervention/resources. The educational objective of this project is to train the next generation of interdisciplinary researchers in the fields of data analytics and healthcare informatics. The progress of the project and the research findings are disseminated via the project website (http://www.cs.wayne.edu/~reddy/projects/health/).

Many real-world phenomena can be modeled by dynamic networks whose connectivity as well as activity changes over time. Hence, there is a growing interest in elucidating the structure and dynamics of such networks. Existing approaches to this problem focus mainly on utilizing either coarse network properties or global structural features to comprehend network dynamics. Such methods often rely on extensions of network features of static networks to understand dynamic networks and fail to capture the rich dynamics of real-world networks.

This exploratory project explores a hierarchical approach to decomposition of network structure and dynamics that can explain changing dynamics at multiple scales ranging from node-level to community-level. The approach is novel, and because of its untested nature, somewhat risky. The research is organized around three aims:(i) Develop information-theoretic flow based approaches that can extract multiple layers of dynamics by simultaneously optimizing for explicit community structures and partial flow dynamics in complex networks. (ii) Develop a computational framework for dynamics-aware network summarization that preserves the flow dynamics of graphs and provides a summary of the large-scale graph dynamics.

The project advances the current state-of-the-art in network data analytics. The resulting tools for elucidating the structure and dynamics of complex networks at multiple scales could potentially transform the way we understand, design, engineer, and control complex networks. The project enriches research-based training and outreach activities at Wayne State University.

This award provides travel support for approximately 26 students to attend the 2013 SIAM Data Mining (SDM 2013) Conference (and the Doctoral Student Forum) that is being organized by the Society for Industrial and Applied Mathematics (SIAM) in cooperation with the American Statistical Association. SDM 2013 is being held in Austin, Texas, USA from May 2 to 4, 2013. The Doctoral Forum will consist of a poster session for Ph.D. candidates in data mining or closely related areas or Ph.D. students who have made significant progress towards Ph.D. candidacy. The Ph.D. student forum participants will be able to present their work, interact with their peers from other universities as well as hundreds of leading researchers in data mining from around the world. In addition, they will attend the technical sessions, plenary talks, and tutorials and workshops of their choice at the conference. Each student will be assigned a senior researcher to serve as a mentor on a variety of career-related issues.

Participation in premier research conferences in data mining is an integral component of the training of Ph.D. students in data mining. The SDM 2013 Doctorul Student Forum is aimed at providing an opportunity for Ph.D. candidates to present their work and receive constructive feedback and mentoring from established researchers in data mining. Such feedback and mentoring is expected to improve the quality of the students thesis research. Similarly, student recipient of a travel award will be able to attend technical sessions, plenary talks, panels, tutorials and workshops. They will interact with peers who share similar interests from other universities, as well as hundreds of leading researchers in data mining from around the world. This experience will be extremely formative and fruitful towards the shaping of their future research endeavors.

Data mining is playing an increasingly important role in many emerging data-rich sciences and application domains, such as healthcare informatics, bioinformatics, computational biology, link analysis, counterterrorism and security. The SDM 2013 Doctoral Student forum will enrich the education and training of student researchers at early stages in their careers. Attendance in SDM 2013 will expose students to cutting-edge research and to relevant applications in a variety of domains. The travel awards will help broaden the participation of women and members of under-represented groups within the Data Mining research community.

DESCRIPTION (provided by applicant): Epithelial-to-Mesenchymal Transition (EMT); a driver of tumor resistance and metastasis, is a complex mechanism that arises through an intricate cross talk between highly robust biological networks. There is minimal information on the most central genes in the networks that drive EMT primarily due to the lack of proper computational tools. To address this unmet problem, two PIs (a computational biologist and a molecular biologist) have teamed together to identify the central genes that are differentially expressed between epithelial and mesenchymal subtypes in well recognized Weinberg's EMT cell models. While these models have been the subject of differentially expressed (DE) gene analyses using the t-test and the F-test, it is not sufficient to interrogate the entire EMT phenomena due to the presence of additional genes that do not meet the DE criteria. Existing models for network analysis, co-expression analysis, and gene clustering can only provide information about a group of genes with similar behavior. However, such analysis cannot extract EMT-specific characterization of mesenchymal pathway genes; i.e. identifying the distinguishing set of mesenchymal patterns in the entire co-expressed gene groups that may be specific to EMT only. Here, we propose a network-based differential analysis model for analyzing the topological differences between two gene networks constructed from the expression data. We hypothesize that for deeper understanding of EMT a differential network analysis coupled with biological validation of the EMT associated genes in the correct models is critical. To this end, we performed comparative genomic microarrays expression investigations on Weinberg's K-ras-HMLE (Epithelial) and K-ras-HMLE-SNAIL (Mesenchymal) 4 cell lines datasets. Our analyses revealed a significant global gene expression difference between parent K-ras-HMLE and HMLE-SNAIL cells. As they are SNAIL driven EMT models, we challenged these cells with a small molecule inhibitor (SMI) against SNAIL (GN-25). Our new computational approach utilizes differential network analysis in multiple EMT models in cell culture, and in animal tumor model (to verify the influence of tumor microenvironment on EMT in situ). This will be coupled with more robust biological validation in the presence of newer network targeted drugs. Therefore, our Specific Aims are 1) Identifying EMT central genes using differential network-based algorithms and 2) Biological validation and evaluation of targeted strategies against centralized genes in the EMT networks. The identified central gene will be validated at the mRNA and protein expression level and their cause-effect relationship will be evaluated using RNA interference in the paired EMT models. In a network-driven drug design, the EMT cells will be challenged with a repertoire of small molecule drugs (identified through our chemical library screening) as single agent or in combination and verify whether drug treatments could target the central genes. Additional validation using efficacy trial of the most potent SMI or its combination in animal tumor models will fortify the clinical application of our network derived EMT targeted drugs.