Working Toward Clinically Meaningful Subtypes of Hypersomnolence Using Data-Driven Phenotyping, but Not There Yet

Abstract

Neurologists establish individualized treatment plans that are based on specific diagnoses. This includes both the main diagnosis (e.g., hypersomnolence) and any subcategories that will influence the neurologist's recommendations (e.g., narcolepsy type 1).1 Traditionally, these diagnoses are based on criteria determined by expert consensus such as the International Classification of Sleep Disorders (ICSD-3).2 While the experts who develop these criteria leverage their extensive experience to justify their clinical value, these criteria are ultimately qualitative and subject to the strengths and limitations of human cognition to identify patterns in complex data. As we enter the era of precision medicine, data-driven methods seek to supplement this human judgment to ensure that these diagnostic criteria truly inform and improve clinical care.