CovidPrognosis

CovidPrognosis predicts COVID-19 patient prognosis using machine learning on clinical parameters and normalized protein expression profiles to identify biomarkers associated with survival or death.

Key Features:

• Machine learning prediction: Employs AI-based prediction algorithms and classification models to predict survival versus death from clinical and protein expression data.
• Performance metrics: The clinical-parameter model achieved 89.47% accuracy, 85.71% sensitivity, and 92.45% specificity, while the protein-expression model achieved 89.01% accuracy, 92.68% sensitivity, and 86% specificity.
• Biomarker identification: Identified 9 clinical and 45 protein-based biomarkers that are associated with patient survival or death and correlate with existing literature.
• Data sources and formats: Uses publicly available datasets comprising clinical parameters and normalized protein expression values from hospital-admitted COVID-19 patients.

Scientific Applications:

• Prognostic stratification: Enables prediction of patient survival risk to support early prognosis and clinical decision research.
• Biomarker discovery and validation: Supports identification and literature-based correlation of clinical and protein biomarkers for COVID-19 disease progression.
• Treatment-outcome research: Provides outcome-associated features that can inform research into therapeutic prioritization and management strategies.

Methodology:

Evaluated AI-based prediction algorithms and classification models on datasets of clinical parameters and normalized protein expression values from hospitalized COVID-19 patients, reporting model performance by accuracy, sensitivity, and specificity and deriving clinical and protein biomarkers from model analyses.