Covid19Risk.ai

Covid19Risk.ai provides a curated repository and executable implementations of externally validated predictive models for early diagnosis and prognosis of COVID-19.

Key Features:

• Comprehensive repository: Aggregates predictive models sourced from peer-reviewed publications and open-source repositories for COVID-19 risk assessment.
• Model validation: Includes models that have undergone external validation reported as TRIPOD type 3 or 2b.
• Model collection: Contains executable implementations of nine predictive models originating from six different research groups.
• Nomogram coefficient extraction: Implements a method to obtain regression coefficients from nomograms when original coefficients are not available.
• Repository extensibility: Designed to incorporate additional externally validated COVID-19 predictive models as they become available.

Scientific Applications:

• Diagnosis: Provides model-based risk estimates to support early diagnosis of COVID-19 in clinical contexts.
• Prognosis and follow-up: Generates patient-specific predictions to inform prognosis, risk stratification, and follow-up planning.

Methodology:

Models were identified via a systematic literature search across PubMed and WHO databases and selected from peer-reviewed publications or open-source repositories; included models report external validation (TRIPOD type 3 or 2b), and regression coefficients are obtained from nomograms when necessary.