Where we are today
Current status: launching Phase I with oncology teams
ARMONEX is currently being developed as part of a PhD in biomedical engineering, following a four‑phase methodology that spans from ontological design and synthetic testing to a prospective multicenter pilot in real time.
The project is now entering Phase I (Modeling), working closely with hospital oncology teams to capture expert reasoning into an executable oncology ontology, validate which toxicities to prioritize, define vital sign thresholds, and agree on the logical rules that will govern the future clinical decision support layer.
Research initiative in clinical validation (H.U. Reina Sofía). IP under protection process.
Phase I – Ontological engineering and clinical design
Co‑designing a dedicated oncology ontology with clinicians: defining key toxicities, vital‑sign‑based triggers, clinical rules, and synthetic stress tests for the explanation middleware.
Phase II – Real‑world data and semantic curation
Integrating and cleaning anonymized historical clinical data, using advanced data quality and clinically constrained imputation techniques to build a robust dataset for model training.
Phase III – Predictive core and explainability
Training temporal AI models to detect early deterioration patterns within a 24‑hour horizon and coupling them with a clinical knowledge graph so that each prediction is paired with a structured, clinician‑friendly explanation.
Phase IV – Silent prospective pilot
Controlled deployment in real clinical practice, with patients monitored at home and a “silent mode” that compares ARMONEX alerts against real‑world outcomes, without interfering with patient management or safety.
Phase V – Scaled clinical adoption
Production rollout of ARMONEX as a regulated, hospital‑grade decision support service, integrated into oncology workflows and EHRs, with alerts actively supporting triage and tele‑oncology.