[GSoC 2026] Project Proposal: Netanel Stern (@NetanelCyber) | Bringing PenuX In Silico Pathogen Analysis to SMART on FHIR

[נתי שטרן‎]

Hello SMART on FHIR Community,

My name is Netanel Stern (@NetanelCyber), and I am a developer specializing in clinical data security and algorithmic analysis. I am writing to introduce my project, PenuX, and to seek guidance on its evolution into a SMART on FHIR application for the GSoC 2026 cycle.

The Project: PenuX

PenuX (github.com/netanelcyber/penuX) is a Pre-clinical Stage I In Silico Only framework. It uses mechanistic modeling to analyze the non-linear correlations between WBC, SpO2, and Temperature to predict the probability of 12 distinct pathogens, with a primary focus on high-risk nosocomial agents like MRSA.

SMART on FHIR Integration Logic

To move this project from a standalone in silico model to a clinical decision support tool, I am proposing the following integration path:

1. Data Ingestion: Utilizing the Observation resource to pull real-time streams of White Blood Cell counts, Oxygen Saturation, and Body Temperature.

2. Pathogen Classification: My algorithm analyzes these vitals to detect specific "Conclusion Signatures." For example, the MRSA gate is triggered by the intersection of thermal spikes ($>39^\circ\text{C}$) and immune paradoxes ($WBC < 4k$ or $> 12k$).

3. Standardized Output: Wrapping the probability results in a RiskAssessment resource to be written back to the EHR, providing clinicians with a "Pathogen Probability Score" alongside the raw data.

4. Security (The Netanel Protocol): Leveraging SMART’s OAuth2 and OpenID Connect protocols to ensure all in silico analysis remains HIPAA/PHI-compliant.

Seeking Mentorship & Feedback

As I prepare my GSoC 2026 application, I am looking for feedback on how to best handle high-frequency vital sign observations within the SMART framework without causing excessive API overhead.

I have already begun exploring the CapabilityStatement of various sandboxes and am eager to contribute to the interoperability of predictive diagnostics.

Respectfully,

Netanel Stern

GitHub: NetanelCyber