Abstract
Background: Data science, rooted in computer science, statistics, and information science is advancing healthcare research by unlocking “insights” from big data to discover knowledge about patient experiences and answer previously unanswerable questions. The comparatively limited engagement of Canadian nurse researchers in this field, relative to counterparts in other jurisdictions (e.g., United States) served as a key impetus for initiating this data science study. Purpose: To investigate homecare electronic health records (EHRs) of “persons with ALS” (PALS) disease and identify care-related factors that influence their preferences and ability to be safely supported at home. Method: Guided by a nursing informatics’ data, information, knowledge, wisdom (DIKW) framework and the knowledge discovery in databases (KDD) methodology, a retrospective, secondary analysis of an integrated dataset (1159 clinical assessments and administrative data) documenting 240 PALS's homecare encounters (April 1, 2009 - July 31, 2019) was conducted. EHR data was analyzed using correlations and logistic regression to model institutionalization risk factors of PALS. Results: Five significant models were generated that accurately distinguished PALS institutionalization status (at home/not home) while offering comparable predictive performance. The final model featuring six factors offers homecare providers real-time insights into PALS clinical status at the point of care. Four relate to ALS clinical manifestations of disease decline and two assessment outcome measures (MAPLe and CHESS), strongly predicting institutionalization and caregiver burden. Conclusion: Big data science offers nurse researchers a transformative way to advance knowledge development that supports high-quality, sustainable healthcare delivery while fulfilling an intended benefit of EHRs.
Funding Information
  • Western University (Ontario Graduate Student Awards)
  • Western University (Western Graduate Scholarships)
  • Registered Nurses’ Foundation of Ontario (Ontario Nursing Informatics Group Award)

This publication has 64 references indexed in Scilit: