Klein Biodefense

What it is

An early-warning system’s job is not to declare an emergency. It is to say something changed, investigate now, early enough that people have time to verify before a decision becomes irreversible.

Klein Biodefense is built for early warning, not for clinical diagnosis. Given the same data and the same configuration it produces the same result every time, so an independent lab, a university or a critic can reproduce it.

What it does

• Multi-pathogen detection. Works on viral, bacterial, toxin, radiological and environmental threat time series.
• Severity staging. Classifies events as LOW, MODERATE, HIGH or SEVERE, for a graduated response.
• Precursor detection. Identifies escalation patterns in the days or hours before a severe event.
• Air-gapped capable. Runs on local data without network or cloud access.
• No neural network. Rule-based and deterministic. No training, no black-box probability, no weights to drift.

Yap Island dengue, 2011–2012

Real recorded data

A retrospective test on 185 daily surveillance observations: real reported case data, not a simulated curve, and not data the system was trained on.

This is same-day detection of the first visible case in a historical dataset. Operational use calls for independent validation under frozen, blinded protocols, and that validation is invited.

Benchmarked real-world events

Real recorded data

Every row is a real event, scored on its real recorded data.

Where it fits

• Edge sensor networks: distributed environmental or public-health monitoring stations
• Command-and-control dashboards: severity tiers map directly to alert thresholds
• Regulatory and public-sector review: a deterministic audit trail, same input, same output

Patent pending (U.S. non-provisional application; priority June 2025).

Read more

• Project details on the Projects page