penuX: In-Silico Pathogen Validation Framework

2 views
Skip to first unread message

נתי שטרן‎

unread,
Jan 9, 2026, 5:30:52 AM (2 days ago) Jan 9
to fhir...@lists.hl7.org, clinic...@ohdsi.org, mimic...@googlegroups.com, hapi...@googlegroups.com, smart-...@googlegroups.com, public-sem...@w3.org, ga4gh-...@googlegroups.com, openmr...@googlegroups.com, dhis...@googlegroups.com, medical-ai...@googlegroups.com, bio-it...@bio-itworld.com, health-it...@hhs.gov, clinic...@googlegroups.com, sepsi...@physionet.org, icu-...@physionet.org

penuX is an open-source clinical decision support (CDS) engine designed to differentiate between 12 critical pathogens through non-linear integration of high-velocity vital signs and laboratory shifts.

🎯 Objective

To provide a standardized, HAPI-FHIR compatible logic layer that identifies "Conclusion Signatures" for infectious agents before definitive culture results are available.

Supported Pathogen Matrix (12 Nodes)

  • Bacterial: MRSA, S. pneumoniae, E. coli, K. pneumoniae, P. aeruginosa.

  • Viral: SARS-CoV-2, Influenza A/B, RSV.

  • Special Cases: Legionella, Mycoplasma pneumoniae, Sepsis (Non-specific).

🛠 Technical Architecture

The framework is built to bridge the gap between raw EHR data and actionable clinical intelligence.

  1. Data Ingestion: Supports FHIR R4 Bundle ingestion via HAPI-FHIR.

  2. Mapping: Normalization of local lab codes to LOINC and SNOMED-CT.

  3. Validation: Logic gates are stress-tested against the MIMIC-IV and eICU retrospective datasets.

  4. Logic Engine: Uses the Netanel Protocol for delta-shift analysis.

The "Netanel Protocol" Logic

The core algorithm calculates the probability of a pathogen-specific signature $P(s)$ based on the velocity of physiological change:

$$P(s) = \int_{t=0}^{n} (\Delta WBC \cdot \Delta Temp) \cdot \frac{1}{\Delta SpO_2} dt$$

  • Bacterial Gate: Identified by a "Thermal Spike" ($>1.5^\circ\text{C}/hr$) + Leukocytosis.

  • Viral Gate: Identified by "Silent Hypoxia" ($SpO_2 < 90\%$) + Stable WBC count + Stable Respiratory Rate.

🚀 Getting Started

Prerequisites

  • Java 17+ (for HAPI-FHIR libraries)

  • Python 3.10+ (for MIMIC-IV feature extraction)

  • Access to PhysioNet (for validation datasets)

Installation

Bash
git clone https://github.com/NetanelCyber/penuX.git
cd penuX
./mvnw install

📅 GSoC 2026 Roadmap

I am actively seeking mentors for Google Summer of Code 2026. My focus areas are:

  • Refining FHIR-to-OMOP CDM mapping for infectious disease profiles.

  • Implementing Digital Twin simulations for in-silico trial acceleration.

  • Optimizing the logic gates for rural healthcare environments (low-resource settings).

🤝 Community & Contact

  • Author: Netanel Stern

  • Location: Rosh Haayin, Israel

  • Mailing Lists: Active on fhir-dev, clinical-omop, and mimic-users.

Reply all
Reply to author
Forward
0 new messages