Profile performance across the defined cognitive dimensions without making clinical claims.
CISTAC 1.0 Theme
AI4Brain
Artificial intelligence meets neuroscience—connecting data-driven brain research with biologically inspired intelligent systems.
CISTAC 1.0 Technical Challenge
Cognitive Growth Agent
Build an AI agent that learns how a person learns, turns performance into an uncertainty-aware profile, and delivers explainable coaching that evolves over time.
01
Preselection deadline
02
Final pitching
Assess. Understand. Coach. Improve.
Create a non-clinical cognitive growth agent that continuously adapts to each user instead of delivering a one-time test result.
Turn scores, response times, errors, and behaviour into a percentage profile that communicates uncertainty.
Give personalised, explainable feedback and motivation in natural language.
Generate adaptive exercises and measurable learning paths that evolve with the user.
The required system
Layers A-C, Brain Twin history, and a documented collaboration-score formula are mandatory.
Assessment & profiling
Design tasks for each cognitive dimension and document definitions, scoring, uncertainty, and possible bias. Collect score, response time, error rate, and behaviour.
Explainable AI coach
Provide clear, personalised coaching in natural language. The architecture is open, but it must be justified and avoid clinical terminology.
Adaptive learning
Adapt exercise difficulty and learning plans. Show progress through radar charts, progress curves, heatmaps, and goal tracking.
Brain Twin & collaboration score
Required: maintain a dynamic, historical user profile and add a documented 0-100 AI Collaboration Score. Future projections are optional and must always show uncertainty.
20% Clarification
25% Critical verification
20% Iteration
20% Autonomy
15% Responsible use
These weights are an unvalidated design proposal; teams must document their chosen formula.
Build for trust
Non-clinical by designDescribe growth and learning without diagnosis or clinical language.
Choice is mandatoryCamera and voice may be used, but a non-biometric mode must always be available.
Show uncertaintyDocument limits and bias. Never present a future projection as certainty.
Respect user dataDocument consent, retention, and deletion. Do not make sensitive decisions from face or voice alone.
What your team delivers
- 01
Public GitHub repositoryYour implementation must be public when submitted.
- 02
Demo videoA focused product demonstration lasting 3-5 minutes.
- 03
Technical reportSubmit a technical report of no more than eight pages.
- 04
PresentationA concise deck of no more than 15 slides.
Two phases. One final defense.
Phase01
Preselection
Submit the public repository, demo video, and presentation. The jury selects the finalist teams.
Phase02
Final pitching
Finalists present for eight minutes and answer questions for five minutes. A surprise element is revealed one day before the defense.
How entries are scored
Each entry is reviewed by at least two evaluators with conflict-of-interest controls.
Phase 1
60%- Relevance20%
- Quality20%
- Innovation20%
- User experience15%
- Explainability15%
- Safety10%
Phase 2
40%- Generalisation25%
- Robustness25%
- Profile quality20%
- Plan relevance15%
- Defense15%
Ready to take part?
Team size
Teams have 2-4 members. Add one member if the team includes an IEEE CIS member, and one more if it includes an IEEE YP member.
Official language
English is required for the challenge, submissions, pitch, and final defense.
Open toolset
Use any external tools, libraries, or frameworks that support your solution.
Your intellectual property
Participants retain full ownership of their work.