Q&A: Unpacking the AU-BISR system

In a world of ever-increasing and varied threats — both natural and human-made — countering dangers posed to public health by dangers to or disruptions of so called “soft targets” such as agriculture, the environment and the nation’s food supply chain are more critical than ever. Yet a system to constantly monitor the safety of such targets is not currently available.

The Auburn University Biosurveillance Intelligence, Surveillance and Reconnaissance System (AU-BISR) aims to address this risk through a multi-tiered system utilizing monitoring, detection, analysis and artificial intelligence (AI) to identify potential chemical, biological and radiological threats in their earliest stages, before they become difficult or more expensive to address. It is garnering widespread government and industry attention, including being prominently featured in a recent edition of Cipher Brief, a major publication in the national security arena.

We spoke with Robert Norton, professor of veterinary infectious diseases, biosecurity and public health in the Auburn University Office of the Senior Vice President for Research and Economic Development, about the system, its potential and its current status.

Briefly, what exactly is the AU-BISR System?

RN: AU-BISR™ is a new process design and infrastructure partnering effort, initiated within the Office of the Senior Vice President for Research and Economic Development of Auburn University. The design uniquely combines existing agricultural, veterinary and biological science with the tactics, techniques and procedures developed over the past few decades to produce actionable intelligence to combat global terrorism and other traditional national security threats to the United States. 

The system consists of a layered sensor array beginning at the microscopic level and extending through its upper layers into space. Sensors communicate with each other in a process called “tipping and cueing.” The technology for sensors and platforms at all the layers currently exists in both the national security (government and military) and private (commercial and academic) sectors. What is novel about AU-BISR is the application of these capabilities against a “non-traditional” target set: the nation’s food, agriculture and water resources and industries, which are vital to national security and the U.S. economy.

What types of scenarios or threats is it designed to counter?

RN: AU-BISR is an agnostic biosurveillance system, meaning the sensors can be used to monitor a variety of situations including the animal health (whether outdoors or in housing), crops, and ecosystems, and public health — including in emergencies such as the aftermath of natural or manmade disasters. Beyond continual monitoring, AU-BISR can also serve a vital role in guiding remediation efforts, such as will eventually be needed in war zones like Ukraine or as is currently needed for monitoring crisis drought areas, such as Iran. 

What shortcomings in current systems would the AU-BISR system address?

RN: Currently, there is no U.S. biosurveillance monitoring system. AU-BISR would be the first such system to comprehensively address the needs included within both biosecurity and national security.  

What about the role of AI in the system, both as an asset and possibly utilized as part of a threat?

RN: AU-BISR will not utilize generative AI, such as provided by Google, but rather use reinforced learning via Artificial Neural Networks (ANN) on closed sets of pre-validated data. This approach eliminates error and prevents AI hallucinations that can occur in generative AI. Before being submitted to the database, all data will be validated to a legal standard for identification. 

For example, an animal pathogen is isolated somewhere in the U.S. At the same time the pathogen is isolated, sensors pass over the affected area, creating a “signature,” an algorithm developed from the findings coming from the totality of sensors used to surveil the area. Once the pathogen is identified according to the legal standards, the data is incorporated into an algorithm that includes both the pathogen ID (e.g., genomic, and other clinical data) and the sensor data, thus creating a “fingerprint.” The findings can then — and only then — be submitted to subject matter experts, confirmed by them as accurate and then incorporated into the training of the AI model. If errors are encountered, the AI model is pulled offline and the error corrected before it is put back online. 

How has the idea of the AU-BISR system been received in government and security circles?

RN: The AU-BISR system is receiving very good feedback from the government. Cris Young (professor of practice, AU College of Veterinary Medicine) and I are currently in discussions with several government and industry decision makers. The [recent] government shutdown has slowed progress, but we hope to hear word on funding in the next few weeks.

If approved, how soon could a fully functional BISR system become operational?

RN: Using commercially available technology, in combination with government technology, would enable a functional prototype (80+% solution) within three years. A fully operational system could be available one to two years after that.