Next-generation safety training that adapts in real-time to individual
trainee performance using machine learning and behavioral analytics.
Instead of one-size-fits-all compliance training, we're developing
systems that learn each person's decision patterns and adjust scenarios
to their specific needs - optimizing learning effectiveness for every
individual.
The Problem for Industry
Nothing can replace interpersonal interaction and training--and no one should try. Nonetheless, supplementary instruction and knowledge transfer
tools face a number of deficits, especially when it comes to remote delivery:
Static scenarios that don't adapt to individual skill levels
Limited behavioral data captured during training
No connection between training performance and on-job risk prediction
The result? Training regimens that don't create behavioral change, and don't reduce incidents
The Problem for Medicine
For individuals who have suffered brain injury or mental trauma, returning to the workforce is a crucial aspect of recovery.
Yet here, too, there are gaps in the tools used to address this problem.
Generalized cognitive exercises: Don't transfer to work tasks
Vocational programs: limited by instructor availability, focus on job matching, less so rehabilitation
The Gap: No adaptive, work-task-based cognitive rehabilitation tools
A new approach
CoAxiom Services is conducting research into a new, AI-led method of instruction. In the system it plans to develop, trainees will participate in an iterative mode of training that leverages
a behavioral feedback loop:
Trainees and patients take a training simulation that emulates real scenarios
During the training, data on trainee actions is recorded and aggregated
After each session, an AI performs analysis on this data and identifies key behavioral trends and construct predictive risk profiles
Parameters within the training simulation are then modified based on this analysis, to encourage desired behavior and discourage undesirable behavior
Trainee behavior in regards to the practiced tasks is thus guided, through this training loop, toward an optimal state
Current Foundation
We have a production browser-based driver safety training platform currently in use that will serve as the bedrock for the project. It currently features
the necessary ingredients to begin building a research prototype:
Realistic 3D driving scenarios with decision points
Cloud infrastructure (Azure) operational and scalable
Performance tracking and behavioral data collection
This production platform provides the foundation for AI/ML integration
and serves as our data collection infrastructure.
Research Partnership Opportunities
We're seeking research partners to co-develop AI-adaptive capabilities:
Universities with human factors/safety research programs
Corporate R&D teams with machine learning expertise
Safety analytics firms with data science capabilities
Insurance research divisions exploring predictive risk modeling
Partnership Structure
CoAxiom:
Training platform, deployment infrastructure,
commercialization pathway
Partner:
AI/ML development, research design, validation,
academic publication
Joint:
Grant applications, intellectual property, research outcomes
What We Bring
Developed cloud infrastructure and data pipeline; grant development experience and writing capability; and path to commercial application
What We're Looking For
Machine learning / AI development expertise
Research design and validation methodology
Federal grant experience (SBIR/STTR preferred)
Publication track record in safety or ML domains
Funding Pathway
NSF SBIR/STTR Program
Phase I: $275,000 (6-9 months) - Feasibility study
Phase II: $1,000,000 - $2,000,000 (24 months) - Development
University, Corporate or Foundation partnership strengthens STTR eligibility
Interested in Collaboration?
If your organization is interested in exploring research partnership
opportunities, we'd love to discuss potential collaboration. Visit our contact page and get in touch!
