Identifying and Typing Latent Neuropathology Using Pattern Recognition on Longitudinal Data

Sean A. P. Clouston , Stony Brook University, State University of New York (SUNY)
Yun Zhang, Stony Brook University, State University of New York (SUNY)

Dementia affected 16 million people and caused 110,000 deaths in the U.S. Dementias result from neuropathology and cause decrements to episodic memory. Alzheimer’s disease and related dementias (ADRD) or by cerebro-vascular and ischemic disease (VaID) cause most dementias, but these are unreliably identified in population data since diagnoses are difficult and expensive. The current study used pattern recognition on longitudinal trajectories of cognitive decline to identify latent, often unreported, cases of VaID and ADRD. The resulting method was reliably able to both identify and differentiate between these two causes of disease. Incidence of VaID and ADRD identified in this way was 40.23/1,000 (95% CI = [39.40-41.08]) and 27.63/1,000 (95% CI = [26.97-28.29]) person-years respectively and was highly concordant with reported diagnoses. Age was significantly associated with higher incidence of both. This was the first study to use pattern recognition to identify and differentially identify latent neuropathology.

See paper

 Presented in Session 5. Health & Mortality 1