A Data-Driven Model to Predict Delirium Based on Dynamic Patterns of Clinical Deterioration in Critically Ill...

Area of Science:

• Critical Care Medicine
• Medical Informatics
• Patient Monitoring

Background:

• Delirium is a common complication in intensive care units (ICUs).
• While many risk factors for delirium are known, the impact of dynamic clinical deterioration patterns is not well understood.

Purpose of the Study:

• To investigate the association between changes in clinical parameters and the risk of developing delirium.
• To identify specific physiological trends that predict delirium onset in ICU patients.

Main Methods:

• Retrospective analysis of electronic health records (EHRs) for 3600 ICU patients.
• Categorization of clinical parameter changes as 'worsen to or persist worsen' or 'recovered to or persist normal'.
• Logistic regression analysis to identify significant risk factors for delirium.

Main Results:

• The predictive model achieved a C-statistic of 0.88, indicating excellent discrimination for delirium development.
• Eight variables showed significant association with delirium, including low diastolic blood pressure, high heart rate, elevated PaCO2, low albumin, elevated blood urea nitrogen, and high sodium levels.
• Persistent or worsening trends in these parameters were linked to increased delirium risk.

Conclusions:

• Worsening or persistent physiological abnormalities are significantly associated with delirium in critically ill patients.
• Monitoring and addressing trends in hemodynamics, acid-base balance, and electrolytes can aid in early delirium prevention.
• Tailored interventions based on identified clinical trends can help prevent delirium in high-risk ICU populations.