Understanding Epilepsy: Big Data Way
Big Data, one of the most sought after complex and voluminous dataset concepts, has a great potential to decipher the unexplored mysteries of the human brain. Owing to the complex origin and extensive workup of patients with refractory epilepsy, we have a unique opportunity to foster innovative approaches that involves mining of diverse datasets comprising of electrical brain signals, neuroimaging, histology and ‘omics.’ Although working on gigantic datasets originating from multiple sources is itself a tedious task, developing an appropriate information platform to store and facilitate research is even more challenging.
Under the guidance of Dr. Jeffrey Loeb, an expert epileptologist and Head of the Department of Neurology and Rehabilitation, I have developed computational signal analysis frameworks to study complex brain networks. Over the past four years, I have worked on investigating the causal propagation of interictal spikes and their interplay with seizure events. Simultaneously, I am also studying the origin and progression of epilepsy in a tetanus toxin injected animal model. With a bio-engineering background and currently working in the Department of Neurology and Rehabilitation, I have a strong inclination to develop predictive computational models with use of machine learning to forecast seizure occurrence from interictal epileptic events. Once completed, the research work will help us understand the brain wirings under pathological conditions such as epilepsy. Moreover, the computational frameworks will aid physicians localizing the disease onset during pre-surgical evaluations.
The other exciting part of my research is developing a data platform for NeuroRepository that stores many complex datasets around a large number of brain disorders. My goal is to design a database to store the forever expanding patient datasets and create an integrated platform to connect different data sources like electronic health records, imaging systems, human tissue samples, and other related data inventories. Once developed, we plan to expand these datasets to different diseases and also link data from different hospital and research centers with stratified levels of permissions to access datasets. We believe this platform would help physicians to store their patient data and expedite their research investigation with simple web infrastructure and intuitive visualization techniques.
Biswajit Maharathi, MS
PhD Y-2, Bioengineering,
Dept. of Neurology and Rehabilitation