- Contribution
- Product, engineering and deployment
- Status
- In use
- Scope
- Medication and dysphagia context · Validated clinical formulary · Clinical retrieval and CIMA/AEMPS · Structured report · Audit and expert review
- Source
- Private
Context
Adapting oral medication for patients with dysphagia is more than deciding whether a tablet can be crushed. Dosage form, type of dysphagia, prescribed IDDSI texture and the available clinical evidence all affect a safe administration plan.
What I built
I translated requirements defined with hospital pharmacists into a working system and implemented medication search through CIMA/AEMPS, a pharmacist-validated clinical formulary, retrieval over indexed clinical documents, structured report generation, automatic IDDSI instructions, audit and expert review, and the deployment workflow. Clinical criteria and final validation remain with the pharmacy professionals.
Evidence and recognition
FarmaGlup is used in a real hospital-pharmacy workflow at Hospital de Manises. In 2026, the hospital team received second prize in Pharma Challenge 6.0, an innovation programme for hospital pharmacy promoted by Bayer.
Public boundary
The software and its operational data remain private. This case study documents confirmed product and engineering decisions without exposing patient information, credentials, production identifiers or internal deployment details.
System overview
- Medication and dysphagia context
- Validated clinical formulary
- Clinical retrieval and CIMA/AEMPS
- Structured report
- Audit and expert review
Three decisions worth discussing
Validated knowledge first
An exact pharmacist-reviewed match is reused before generative retrieval is considered.
Traceable recommendations
Every result identifies its evidence path and the original generated report remains available for audit.
Structured clinical output
Typed report fields constrain safety signals, instructions, alternatives and bibliography instead of relying on free text.
From problem to evidence
- 1
ProblemReviewing oral medication for dysphagia requires reconciling dosage form, swallowing condition, prescribed texture and clinical evidence.
- 2
ConstraintGenerative output cannot replace clinical validation and every recommendation must remain traceable to its source.
- 3
DecisionUse a validated formulary first, then clinical-document retrieval and CIMA/AEMPS fallback, producing a structured auditable report.
- 4
EvidenceThe system is used in a hospital-pharmacy workflow and received second prize in Pharma Challenge 6.0.
Technical notes
Layered evidence pathOpen +
The system checks an exact pharmacist-validated formulary entry first, retrieves indexed clinical documentation when needed and uses CIMA/AEMPS as a bounded fallback.
Structured safety reportOpen +
A typed schema controls traffic-light classification, administration instructions, warnings, alternatives and bibliography before the interface renders the result.
Audit and expert feedbackOpen +
Original consultations remain as evidence while pharmacist corrections create or update reviewed formulary entries with their own history.
Current limitations
- The system supports rather than replaces clinical judgement; every recommendation requires pharmacist validation.
- Broader institutional use requires stronger identity, permissions and clinical regression coverage.