Mohammad Adibuzzaman

Mohammad Adibuzzaman

Co-Director, Informatics, @OCTRI, OHSU

LATICE: Framework for Longitudinal Analysis of Time-Course Data from Patient Generated Health Data

September 10, 2019

Talk, Discovery Park Convergence Conference 2019, Discovery Park Purdue University, West Lafayette, Indiana


We aim to develop critical and enabling cyberinfrastructure for integrated analyses of longitudinal and time-series data from the multitude of sensors in common commercial wearables, in conjunction with the corresponding patient pharmacological and electronic health records EHRs). Effective analyses of this rich source of data has the potential to significantly enhance long-term individual well-being, reduce cost of healthcare, improve our understanding of pathology, associated sensor markers and prognoses, characterize the efficacy of drugs and identify adverse effects and interactions, and to enable a broad class of new data-driven studies. Architecting the proposed system poses significant challenges stemming from the heterogeneity of sensor devices and associated quality of data, diversity of populations and underlying pathologies, disparate drug regimes and responses, and complexity of the underlying analytics problems.

An Open-access High Performance Computing System for Developing Research Applications (Apps)

August 17, 2017

Talk, AMIA Webinar, West Lafayette, Indiana

Despite the promise of big data, little evidence has been generated for clinical practice with data driven systems. A new model for collaborative access, exploration, and analyses of integrated clinical data will be presented with a standard database, Medical Information Mart for Intensive Care - III (MIMIC III), for translational clinical research. The proposed model addresses the significant disconnect between data collection at the point of care and translational clinical research. It addresses problems of data integration, pre-processing, normalization, analyses (along with associated compute back-end), and visualization. The pre-packaged analyses toolkit is easily extensible, and allows for multi-language support. The platform can be easily federated, mirrored at other locations, and supports a RESTful API for service composition and scaling.

Closing the Data Loop: An Integrated Open Access Analysis Platform for the MIMIC Database

September 12, 2016

Talk, Computing in Cardiology (CinC 2016), Vancouver, Canada

We describe a new model for collaborative access, exploration, and analyses of the Medical Information Mart for Intensive Care - III (MIMIC III) database for translational clinical research. The proposed model addresses the significant disconnect between data collection at the point of care and translational clinical research. It addresses problems of data integration, preprocessing, normalization, analyses (along with associated compute back-end), and visualization. The proposed platform is general, and can be easily adapted to other databases. The pre-packaged analyses toolkit is easily extensible, and allows for multi-language support. The platform can be easily federated, mirrored at other locations, and supports a RESTful API for service composition and scaling.