Investigating NeuroAI: The Journey Towards Understanding Brain-Driven AI

Introduction to the Investigation Phase

As part of our NeuroAI project, we are following the Challenge-Based Learning (CBL) methodology, where each phase builds on the previous one to develop a well-structured solution. Currently, we are in the Investigation Phase, where our main goal is to explore and understand the key concepts related to our project. Through structured discussions, expert insights, and collaborative research, we have delved deep into the intersection of neuroscience and artificial intelligence.

This blog post summarizes our findings from Meetings 6 and 7, covering the guiding questions, guiding activities, and investigation synthesis that have shaped our project so far.

Guiding Questions

To ensure a focused approach in our investigation, we formulated essential questions that would guide our research:

1. What are the concerns of professionals regarding AI in mental health?

2. How can AI enhance the interaction between psychiatrists and patients?

3. What types of AI-driven solutions already exist in the field of neuroscience and psychology?

4. How can we interpret human emotions and cognitive states?

5. What datasets are available for training AI models in neuro-driven applications?

6. How can our solution be implemented in real-world psychiatric sessions?

   
   

These questions set the foundation for our investigation and helped us seek relevant information from professionals and existing research.

Guiding Activities

To answer our guiding questions, we conducted a variety of activities that allowed us to gather insights and refine our understanding:

1. Expert Consultation

One of the most valuable activities was a meeting with the university's psychiatrist. During this discussion, we gained crucial insights into the challenges professionals face with AI applications in mental health.

• Key concern: The psychiatrist was skeptical about AI, particularly due to its lack of 100% accuracy.

• Potential use case: AI could serve as a supplementary tool to aid psychiatrists rather than replace them entirely.

• Psychiatric Techniques Discussed: Brief therapy and psychoanalysis were explored as potential areas where AI could play a role in supporting diagnosis and treatment.

2. Research on EEG and AI

To understand how brain activity can be translated into meaningful insights, we focused on EEG (Electroencephalography) technology and its applications.

• We explored EEG-based sentiment analysis to determine emotions and mental states.

• We investigated image reconstruction from EEG data, a technique where brain waves can be translated into images.

• We reviewed datasets such as EEG-MNIST and ImageNet, which provide EEG data collected from individuals interacting with various visual stimuli.

A snippet of our research to understand the different meanings of frequencies in the human's mind:

The 10–20 system or International 10–20 system is an internationally recognized method to describe and apply the location of scalp electrodes in the context of an EEG exam, polysomnograph sleep study, or voluntary lab research. This method was developed to maintain standardized testing methods ensuring that a subject's study outcomes (clinical or research) could be compiled, reproduced, and effectively analyzed and compared using the scientific method. The system is based on the relationship between the location of an electrode and the underlying area of the brain, specifically the cerebral cortex.

Below is an image illustrating the 10-20 system:

3. Brainstorming and Knowledge Sharing

During our meetings, each team member contributed research findings, relevant datasets, and case studies:

• Bassem shared YouTube videos explaining EEG and its applications in AI.

• Hajer and Sabaa reiterated the psychiatrist’s concerns and highlighted key moments where AI could assist in therapy sessions.

• Collaborative discussions helped us define possible implementations of our solution and refine our approach.

Investigation Synthesis

After thorough discussions and research, we arrived at key conclusions that will shape the next steps of our project:

1. When Should the Solution Be Used?

Our investigation confirmed that our solution should assist psychiatrists in two scenarios:

• At the beginning of a session to facilitate communication and understanding.

• During ongoing therapy to provide deeper insights and prevent communication deadlocks between the psychiatrist and patient.

2. How Can Our Solution Be Implemented?

Through brainstorming, we identified three possible AI-driven implementations:

• Sentimental/emotional analysis through EEG to assess a patient’s mental state.

• Image reconstruction through EEG to visualize patient thoughts and emotions.

• (Further exploration needed for a third method)

3. Addressing AI Skepticism in Mental Health

Given the psychiatrist’s concerns, we recognize the need for:

• A transparent AI system with explainable results.

• A solution that complements psychiatrists rather than replaces them.

• A high-accuracy approach that minimizes potential risks and enhances professional trust.

Related PVs

In the below docs link, check the meeting summaries 6 and 7 related to this phase:

https://docs.google.com/document/d/1STzaunNOn6xlyQjVtN9BcVp-MwhMXvwM-2-TlhX1vds/edit?usp=sharing

Next Steps

With a clearer understanding of the investigation phase, our next meeting will focus on refining our implementation strategies and sharing individual research findings. Our journey in NeuroAI continues as we transition into deeper exploration and development.

Stay tuned for more updates on our progress!