Accurately diagnosing epilepsy is very challenging and time consuming because clinicians rarely observe the actual seizure, and there are many different types of seizures and epilepsy syndromes with differing presentations. Furthermore, other neurological disorders can be mimics of seizures leading to erroneous diagnosis, inappropriate treatments with significant potential adverse events, incorrect prognoses, and significant waste of healthcare resources. A primary indication that is mistaken for epilepsy is psychogenic non-epileptic seizure (PNES) which is commonly misdiagnosed even by seasoned clinicians. Patients that suffer PNES events experience convulsive events which are similar to epileptic seizures but do not arise from aberrant, synchronous, electrical activity in the brain. Rendering a definitive diagnosis of either PNES or epilepsy is critical to long-term patient health and outcome. In many cases, patients suffering PNES events are prescribed anti-epilepsy drugs (AEDs) and suffer unnecessary side effects. Alternatively, patients that are suspected to suffer PNES, but actually have epilepsy and are not treated with AEDs, are at risk of experiencing seizures and even sudden unexplained death in epilepsy (SUDEP). Currently, obtaining a definitive diagnosis of epilepsy or PNES is expensive and inconvenient for patients as it may require inpatient evaluation and a battery of costly tests. Thus, a critical gap in our clinical assessment of seizures, be it epileptic or psychogenic, is an accurate diagnostic blood test that can be used to stratify transient neurological events as either PNES or epilepsy. Towards developing this test, Evogen has developed an innovative informatic approach that incorporates all available clinical information and our protein biomarker approach to detect seizure using an artificial intelligence approach of machine learning. To confirm our previous results, Evogen proposes an epilepsy monitoring unit-based all-comers study which will collect both blood samples and clinical information from patientsÂ’ ES and PNES events. Protein biomarker concentrations will be determined and combined with clinical information using a machine learning approach to create a diagnostic algorithm that will provide a probabilistic score that a patient experienced an ES or PNES event. The test will have a profound impact on diagnosis and treatment of both ES and PNES. We believe this test will have an immediate impact on identifying which patients require EMU evaluation. Typically, one third of patients that visit the EMU for a definitive diagnosis suffer from PNES events and represent a large burden to the tertiary epilepsy centers. When patients are referred for EMU monitoring, they are sequentially put on a waiting list and infrequently is any triage used. Therefore, a test that can aid in routing patients with psychogenic events to the appropriate care will increase the accessibility for patients with a high probability of epilepsy. The work proposed here is the first step necessary to confirm pilot results. Upon confirmation, a streamlined development plan has been created to collect all of the information required for CLIA accreditation and bring the test to market.
Public Health Relevance Statement: Project Narrative Differentiating epileptic seizures from psychogenic nonepileptic seizure is a substantially unmet medical need. Current tools to differentiate between epileptic seizures and psychogenic nonepileptic seizures have limited performance, high costs and limited availability. A diagnostic test that indicates if a patient recently experienced an epileptic seizure or psychogenic event would provide physicians with objective, actionable data that can result in improved clinical care, patient outcomes and healthcare system savings.
NIH Spending Category: Brain Disorders; Clinical Research; Epilepsy; Machine Learning and Artificial Intelligence; Neurodegenerative; Neurosciences; Patient Safety; Prevention
Project Terms: accurate diagnosis; Adoption; Adverse event; Algorithms; Antiepileptic Agents; Artificial Intelligence; base; Biological Assay; Biological Markers; Blood; Blood specimen; Blood Tests; Brain; Caring; CCL8 gene; Cessation of life; CLIA certified; Clinic; Clinical; Clinical assessments; clinical care; Clinical Research; clinical risk; cohort; Collaborations; Confidence Intervals; Consumption; cost; Data; design; Development; Development Plans; Diagnosis; Diagnostic; Diagnostic tests; Electroencephalography; Enrollment; Enzyme-Linked Immunosorbent Assay; Epilepsy; Evaluation; Event; experience; Future; Health; Healthcare; Healthcare Systems; improved; Industry Standard; Inflammation; Informatics; innovation; Inpatients; Intercellular adhesion molecule 1; Kolmogorov-Smirnov Test; Laboratories; Machine Learning; Medical; Methodology; Monitor; nervous system disorder; Neurologic; Outcome; Patient-Focused Outcomes; Patients; Pennsylvania; Performance; Peripheral; Pharmaceutical Preparations; Physicians; Pilot Projects; Plasma Proteins; Population; Probability; Production; protein biomarkers; Proteins; Proteomics; prototype; Recording of previous events; research clinical testing; Resources; Risk; Risk Factors; Route; Running; Sampling; Savings; Seasons; Seizures; Sensitivity and Specificity; side effect; Signal Transduction; Specificity; statistics; success; Surveys; Symptoms; Syndrome; Testing; Time; TNF gene; TNFSF10 gene; tool; Triage; Universities; Validation; validation studies; Visit; Waiting Lists; wasting; Work
