Integrating data analytics into health informatics: Advancing equity, pharmaceutical outcomes, and public health decision-making

Abstract

Introduction: The integration of data analytics into health informatics has become vital for transforming raw clinical information into actionable insights that improve patient care and pharmaceutical outcomes. Objectives: This study uses the Medical Information Mart for Intensive Care IV (MIMIC-IV) electronic health record dataset to examine differences in pharmaceutical prescription patterns and their relationship to clinical outcomes. We investigate how demographic characteristics, including age, gender, and race, affect prescribing patterns for three major drug classes: opioids, antibiotics, and antipsychotics. Methods: We analyzed the MIMIC-IV intensive care unit dataset, incorporating preprocessing of demographic and prescription data to support fairness and outcome analysis. A decision tree model was trained to predict in-hospital mortality and evaluated using standard performance metrics. Results: We examined the relationship between drug type and patient outcomes, finding that antibiotic prescriptions were associated with shorter hospital stays, whereas antipsychotic prescriptions were linked to longer hospitalizations. Our findings reveal statistically significant differences in prescribing patterns, where men were more likely to receive opioids, whereas women were more likely to receive antibiotics. In addition, considerable racial disparities suggest possible systemic inequities. Nevertheless, there was no statistically significant correlation between drug type and in-hospital mortality, indicating that underlying clinical conditions may play a more substantial role. The model achieved an area under the receiver operating characteristic curve of 0.9337 and an F1-score of 0.8235, outperforming several complex algorithms whereas remaining easily interpretable—an important advantage in clinical practice. Conclusion: These results demonstrate the potential of transparent machine learning models to support enhanced medical decision-making and highlight the need for prescription strategies that prioritize fairness and equity.

Ähnliche Arbeiten