• Epilepsy Research

    According to the national average, it takes 10 years to be referred to an epilepsy surgical clinic. At Cincinnati Children’s it takes an average of 6 years. While we are better than the national average, we know we can do better.

    Our lab is working on using the latest advances in artificial intelligence to decrease the time for surgical referral to just 3 years.

  • The Childhood Absence Epilepsy (CAE) study is a randomized, double blind, comparative trial of ethosuximide (ETX), lamotrigine (LTG) and valproic acid (VPA) as initial monotherapy for children with CAE. The primary objective is to identify the optimal anticonvulsant (i.e. the AED with highest rate of seizure control and lowest incidence of treatment limiting toxicity) used for initial treatment based on short-term and long-term outcomes.

    We are contributing to the work of the Comprehensive Epilepsy Center by building algorithms that will automate the summary of large amounts of patient information. Patients with seizure disorders – particularly those with complex seizure disorders – are evaluated with a variety of diagnostic tests and tools whose results vary in complexity, number and size. Our goal is to build machine-learning algorithms that are able to assist clinicians in quickly and efficiently summarizing these reports for the numerous patients treated at the Comprehensive Epilepsy Center.

    We are working on a technology that will use epilepsy free-text progress notes to extract relevant intractable information that will be fed into a decision support tool. In our preliminary studies we found the tool is indeed capable of recommending more surgeries for patients than the clinical staff does. Obviously, the final approval for surgery Phase I evaluation is done by an expert. But even this simple improvement puts fewer burdens on clinicians due to narrowing down the pool of patients that need full retrospective chart reviews.

    We proposed a refractory status epilepticus (RSE) registry that will enable us to determine the most effective treatment and monitoring algorithms by establishing a cohort of well-identified RSE patients to identify biomarkers of outcome. Within the pSERG network we will develop a prospective RSE registry among tertiary care hospitals focused on standardized RSE outcome assessment. This will include the expansion of a standardized web-based data entry tool, development of an outcome registry, and determination of illness severity adjusters and outcome predictors in RE as well as analysis of genetic biomarkers.

    The purpose of this study is to develop a series of epilepsy measurement indicators, automatically collect the data required for these indicators and implement the Multi-Institutional Pediatric Decision Support System (MiPEDS) system, as the basic technology layer for anonymizing data for conducting cross-institutional comparisons.

    The MiPEDS Project utilizes portions of the i2b2 SHRINE framework, and the CHRISTINE system. The i2b2 center provides access to an informatics framework, which allows researchers to utilize existing clinical data for new research. It captures data directly from the Electronic Health Center (EHC), and other sources. This data is integrated into a larger repository, thus allowing data to be analyzed without jeopardizing patient privacy. I2B2 SHRINE links local data into a multi-institutional network, allowing researchers to obtain information from a larger patient set.

    Project Goals:

    1. Achieve expert consensus about which pediatric epilepsy measures are optimal for cross-institutional comparison.
    2. Develop a federated epilepsy data mart that is comprised of the free-text and fixed data required for computing those measures.
    3. Graphically present the measures by using traditional analytical tools and natural language processing methods.

    Medical researchers have "...the need to analyze large amounts of data in an interactive manner quickly with no opportunity to rely on the existence of canned queries...." * A star schema design provides structures that enable faster interactive analyzing of data. "The star data model [is] a design that makes slicing and dicing one specific subject area easy and fast."* An i2b2 data mart is best suited for tasks like cohort identification, hypothesis generation and retrospective data analysis.

    It is NOT well suited for tasks like clinical trials, sample tracking, study administration or providing real-time alerts.

    *"A Layman's Understanding of Star Schemas" by Kevin Meade, 2009-09-26.

    The i2b2 data mart is a data warehouse modeled on the star schema structure. It contains fact tables, to house facts about patients. It also contains tables that provide additional information about fields in the fact table, called dimension tables. Facts are defined by concept codes. The hierarchical structure of these codes, together with their descriptive terms and some other information, forms the i2b2 ontology (aka: metadata).


    CHRISTINE (Children’s Hospital Resource ISelecting Therapy Individualized Expert) is a prototype application which provides decision support for epilepsy patient drug selection. Epilepsy patient data is collected in the CHRISTINE application along with a drug-drug interaction database, and expert input. A drug selection report can be generated based on this data to provide the top 5 best drug options to support any decisions made by the clinician.

    The CHRISTINE application code has spawned a common web application framework for all applications built by the Pestian Lab. So decision support application CHRISTINE is one application among 10+ other applications using a common web application framework which we also call “the CHRISTINE framework”.