Pediatric Workflow Prototype
Patient-Centric Pediatric Workflow Innovation
Structured Records
Mapped pediatric workflows using structured digital patient information.
ML Categorization
Prototype classified medical letters into clinical categories.
Clinical Insights
Revealed strong potential for AI-supported pediatric workflows.
About
Description
MAIA was a conceptual research initiative aimed at improving how pediatric patient records are processed, structured, and used in clinical environments. The project explored how digital workflows could reduce friction for physicians and enhance clarity in cases involving children, without triggering MDR classification.
Scope of the Project
- Definition of pediatric patient workflows and record-processing structures
- Exploration of more intuitive clinician–patient interactions
- Categorization of medical letters into clinical specialties (e.g., pulmonology, gastroenterology)
- Early AI-assisted concepts tested using Jupyter notebooks
- Examination of feasibility, accuracy, and clinical applicability
HEPIUS Responsibilities
- Acting as technical advisor throughout the conceptualization phase
- Designing early prototypes to demonstrate improved pediatric record handling
- Coordinating an 18-month collaboration with FH Technikum Wien students for ML evaluation
- Outlining a future-ready technical architecture using AWS and Flutter
- Assessing compliance boundaries and practical opportunities for clinical integration
Technology Components
- Early predictive categorization using NLP prototypes
- Jupyter notebooks for ML experimentation
- AWS healthcare services (e.g., HealthLake) for structural evaluation
- Flutter for prototype UI concepts
- Academic collaboration to support research validity
Results
Structured Records
- Defined a structured view of pediatric patient journeys, highlighting efficiency gains for physicians.
- Introduced repeatable data flows enhancing clarity in complex pediatric cases.
- Established groundwork for future clinical digitalization initiatives.
ML Categorization
- Prototype models successfully classified medical letters into specialty groups.
- Identified accuracy thresholds and dataset requirements for future production usage.
- Raised foundational insights into ML feasibility within pediatric contexts.
Clinical Insights
- Revealed strong potential for AI-supported pediatric workflows that preserve clinician control.
- Clarified acceptance criteria, accuracy expectations, and regulatory boundaries.
- Provided stakeholders with a clear landscape of opportunities and risks.
Pilot Outcome
- Stakeholders expressed strong interest in continuing the concept, but the emergence of ChatGPT and evolving legal requirements (MDR, AI Act) shifted industry focus.
- The project concluded with a comprehensive insight package, supporting future planning while reducing uncertainty for all participants.
- Although not advanced to production, MAIA remains a valuable reference for responsible innovation in healthcare.
