Please enable JavaScript.

Coggle requires JavaScript to display documents.

AI in pharma, Screenshot 2025-03-24 at 12.02.19 PM - Coggle Diagram

- - - - Resource
      - Description
      - Patient
      - Demographic info (name, DOB, etc.)
      - Observation
      - Lab results, vital signs, metrics
      - Encounter
      - A hospital visit or consultation
      - Medication
      - Drug prescriptions
      - Condition
      - Diagnoses like diabetes, asthma
  - - - Data Ownership & Stewardship
      - Assigns accountability for data assets. Data Owners define how data should be managed, and Data Stewards ensure itâ€™s accurate and properly maintained.
      - Data Quality Management
      - Ensures data is accurate, complete, timely, and consistent. Includes profiling, validation, cleansing, and monitoring processes.
      - Metadata Management
      - Maintains information about the data â€” such as definitions, origins, formats, and relationships. This is critical for data discovery, lineage, and usage.
      - Data Cataloging
      - Tools and practices to organize and surface available data assets, helping users find and understand what data exists.
      - Data Lineage
      - Tracks the flow of data from source to destination (ETL pipelines, transformations, etc.) This is crucial for transparency and debugging.
      - Data Classification
      - Labels data based on sensitivity (e.g., public, internal, confidential, PII/PHI). Helps enforce proper handling and protection policies.
      - Data Security & Access Control
      - Defines who can access what data and under what circumstances, using roles, policies, and encryption.
      - Regulatory Compliance
      - Ensures data practices align with external regulations like HIPAA, GDPR, CCPA, SOC2, etc.
      - Policies, Standards & Procedures
      - Formal rules for how data should be created, stored, shared, and deleted. Helps standardize behavior across teams.
    - - Model Lifecycle Management
        
        Tracks the entire journey: from development, validation, deployment, monitoring, to retirement.
        
        Version Control
        
        Keeps a history of model versions, including training data, code, hyperparameters, and evaluation metrics.
        
        Auditability
        
        Enables tracing model outputs back to the version, data, and logic used â€” critical for regulatory audits.
        
        Explainability
        
        Documents how the model works, and explains its predictions (especially for black-box models).
        
        Bias & Fairness Audits
        
        Identifies and mitigates biases (e.g., racial, gender) in model inputs and predictions.
    - - Data Catalog AWS Glue Data Catalog
      - Access Control AWS IAM, Lake Formation
      - Sensitive Data Detection Amazon Macie
      - Metadata Management AWS Glue, Amazon DataZone
      - Data Lineage AWS Glue + Amazon CloudWatch Logs