A0240
Title: Precision nosology for mental health research
Authors: Thaddeus Tarpey - New York University (United States) [presenting]
Abstract: Nosology is the branch of medical science dealing with the classification of diseases. Nosological research has focused on prototypical approaches postulating the existence of distinct disease categories and dimensional approaches where disease severity varies along a continuum with no clear demarcating disease boundaries. In some mental health diseases, such as depression and ADHD, disease categories have become more and more subdivided across the distribution of diagnostic features to accommodate heterogeneity within reified disease categories. This refining of psychopathology points to the need for precision nosology to individualize diagnoses and treatment based on patient-specific features across data modalities. This is clearly a problem where statistical modeling can play an important role. Common statistical approaches have used tools such as factor analysis and clustering. The use of independent component analysis is proposed to obtain flexible nonparametric density estimates in high-dimensional feature spaces that can accommodate skewing effects due to psychopathology. This approach allows for a nosologically-based projection pursuit modeling to help discover lower-dimensional subspaces to guide the discovery of disease heterogeneity. Approaches that can incorporate data from multiple modalities will also be discussed.
