BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Phoenix
X-LIC-LOCATION:America/Phoenix
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19700101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20241014T203101Z
LOCATION:Grand Ballroom
DTSTART;TZID=America/Phoenix:20240910T154000
DTEND;TZID=America/Phoenix:20240910T160000
UID:HFESAM_ASPIRE - Presented by HFES_sess157_LECT243@linklings.com
SUMMARY:Evaluating Active Learning Strategies for Automated Classification
  of Patient Safety Event Reports in Hospitals
DESCRIPTION:Lecture\n\nShehnaz Islam and Myrtede Alfred (University of Tor
 onto), Dulaney Wilson (Medical University of South Carolina), and Eldan Co
 hen (University of Toronto)\n\nPatient safety event (PSE) reports, documen
 ting incidents that compromise patient safety, are fundamental for improvi
 ng healthcare quality and safety. Accurate classification of these reports
  is crucial for analyzing trends, guiding interventions, and organizationa
 l learning. However, this process is labor-intensive due to the high volum
 e and complex taxonomy of reports. Previous work has shown that machine le
 arning (ML) can automate PSE report classification; however, its success d
 epends on large manually-labeled datasets. This study leverages Active Lea
 rning (AL) strategies with human expertise to streamline PSE-report labeli
 ng. We utilize pool-based AL sampling to selectively query reports for hum
 an annotation, developing a robust dataset for training ML classifiers. Ou
 r experiments demonstrate that AL significantly outperforms random samplin
 g in accuracy across various text representations, reducing the need for l
 abeled samples by 25-70%. Based on these findings, we suggest that incorpo
 rating AL strategies into PSE-report labelling can effectively reduce manu
 al workload while maintaining high classification accuracy.\n\nTrack: Heal
 th Care\n\nSession Chairs: Maryam Tabibzadeh (California State University,
  Northridge) and Shababa Matin (Rice University)
END:VEVENT
END:VCALENDAR