BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//NextToU//EN
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:797744ef-cc7b-4e10-acc8-3d9da43d10b7@nexttoyou
DTSTAMP:20261105T152100Z
DTSTART:20261106T170000Z
DTEND:20261106T180000Z
SUMMARY:Data Science and Statistics Seminar by Dr. Parker Knight (UNC Chap
 el Hill)
LOCATION:Engineering and Computer Science South (ECSS) 2.203\, 800 W. Camp
 bell Road\, Richardson\, Texas 75080-3021
DESCRIPTION:Title: Latent Factor Point Processes for Patient Representatio
 n in Electronic Health Records\n\nAbstract: Electronic health records (EH
 R) contain valuable longitudinal patient-level information\, yet most sta
 tistical methods reduce the irregular timing of EHR codes into simple cou
 nts\, thereby discarding rich temporal structure. Existing temporal model
 s often impose restrictive parametric assumptions or are tailored to code
  level rather than patient-level tasks. We propose the latent factor poin
 t process model\, which represents code occurrences as a high-dimensional
  point process whose conditional intensity is driven by a low dimensional
  latent Poisson process. This low-rank structure reflects the clinical re
 ality that thousands of codes are governed by a small number of underlyin
 g disease processes\, while enabling statistically efficient estimation i
 n high dimensions. Building on this model\, we introduce the Fourier-Eige
 n embedding\, a patient representation constructed from the spectral dens
 ity matrix of the observed process. We establish theoretical guarantees s
 howing that these embeddings efficiently capture subgroup-specific tempor
 al patterns for downstream classification and clustering. Simulations and
  an application to an Alzheimer's disease EHR cohort demonstrate the prac
 tical advantages of our approach in uncovering clinically meaningful hete
 rogeneity.\n\nhttp://next-to-u.app/events/797744ef-cc7b-4e10-acc8-3d9da43
 d10b7
URL:http://next-to-u.app/events/797744ef-cc7b-4e10-acc8-3d9da43d10b7
END:VEVENT
END:VCALENDAR