Data Preparation for Supervised Learning: Improving Nursing Situation Awareness to Reduce Healthcare-Acquired Urinary...

Area of Science:

• Nursing Informatics
• Patient Safety Research
• Clinical Data Science

Background:

• Situation awareness (SA) is a critical non-technical skill for nurses, directly impacting patient care and safety.
• Enhanced SA in nurses can lead to earlier detection of clinical changes, thereby preventing patient harm.
• Healthcare-Acquired Urinary Tract Infections (HAUTI) represent a significant clinical challenge where improved nursing SA could be beneficial.

Purpose of the Study:

• To explore the utility of nursing assessment data from Electronic Health Records (EHRs) for predicting Healthcare-Acquired Urinary Tract Infections (HAUTI).
• To detail the data preparation methods for applying supervised learning algorithms to nursing assessment data for HAUTI risk prediction.
• To identify and discuss challenges associated with data missingness in EHRs for predictive modeling.

Main Methods:

• Extraction of nursing assessment data from Electronic Health Records (EHRs).
• Data preprocessing techniques to prepare the extracted data for supervised machine learning.
• Development of supervised learning models for the prediction of HAUTI.

Main Results:

• The study successfully prepared nursing assessment data for supervised learning.
• Methods for data extraction and preparation were established for HAUTI prediction.
• Significant challenges related to data missingness in EHRs were identified and discussed.

Conclusions:

• Nursing assessment data within EHRs holds potential for developing predictive models for HAUTI.
• Effective data preparation is crucial for leveraging EHR data in supervised learning for patient safety initiatives.
• Addressing data missingness is a key challenge in utilizing EHR data for clinical risk prediction.