Data Analysis Design of Key Indicators in a Machine Learning-Based Intelligent Healthcare System

Abstract

Based on an analysis of real-world patient data, this study conducts exploratory data analysis to identify key risk factors associated with prolonged hospital stays (>7 days) in the context of developing a machine learning-based intelligent healthcare system. Using a dataset of 1,000 anonymized patient records, the research employs Python to examine the influence of age, BMI, blood pressure, and cholesterol levels on hospitalization days. The results reveal that high-risk patients constitute 41.3% of the total population, with significant variations across demographic and clinical categories. Notably, underweight patients (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$B M I&lt;18.5$</tex>) and those with hypotension or prehypertension exhibit markedly elevated risks. Age demonstrates a Ushaped risk relationship, with the youngest (<25 years) and middle-aged (46-55 years) cohorts showing the highest prolonged stay rates. Cholesterol levels, however, show no clear association with extended hospitalization. The study underscores the nonlinear and complex relationships between common health metrics and hospitalization outcomes, emphasizing the role of physiological extremes as markers of underlying frailty and acute illness. These findings provide a foundational basis for future machine learning models aimed at predictive risk assessment and personalized clinical decision support.

Ähnliche Arbeiten