DESIGN OF AN ENTITY–RELATIONSHIP MODEL FOR AN AI-ASSISTED CARDIOVASCULAR DISEASE DIAGNOSTIC INFORMATION SYSTEM

Background: Cardiovascular diseases remain the leading cause of mortality worldwide, and early diagnostic support requires not only predictive algorithms but also reliable clinical data infrastructure. Objective: This study proposes a conceptual and logical database design based on the Entity–Relationship (ER) model for an AI-assisted cardiovascular disease diagnostic information system. Methods: The database model was developed through requirement analysis, entity identification, attribute specification, relationship mapping, key definition, normalization, and relational transformation. The proposed structure integrates patient demographics, medical history, clinical visits, vital signs, laboratory results, electrocardiogram records, AI-based risk assessments, treatment plans, user roles, and audit logs. A proof-of-concept evaluation was performed using a 500-record synthetic test database to assess data consistency, redundancy reduction, query readiness, and technical compatibility with a machine-learning diagnostic workflow. Results: The proposed ER model separates clinical objects into normalized entities and defines explicit one-to-many and one-to-one relationships among patient records, examinations, diagnostic inputs, AI outputs, and treatment recommendations. Compared with an initial flat schema, normalization reduced duplicated fields by 27% in the prototype database. In the synthetic workflow test, the AI module achieved 92.0% accuracy, 92.2% precision, 92.2% recall, 91.8% specificity, and a 92.2% F1-score; these figures demonstrate technical feasibility rather than clinical efficacy. Conclusion: The proposed ER model provides a structured foundation for AI-assisted cardiovascular diagnostic systems by improving data integrity, traceability, security, and readiness for machine-learning analysis. Future work should validate the model using ethically approved, anonymized real-world clinical datasets.

Перевод пока недоступен