Participants: Speaker 1 (Mentor/University Representative), Speaker 2 (PM), Speaker 3 (Team Member)

Main Topic: Discussion of next steps, customer interaction, legal aspects, and development strategy

1. Technical Implementation: The Role of LLM

  • Concept clarification: The team discussed how the patient simulation will work. The LLM will generate patient responses based on context (symptoms, age, medical history) provided by the selected clinical case.

  • Workflow: The student (doctor) sees the context, talks to the LLM, orders tests (the system returns predefined results for that case), and finally fills out a card to compare their diagnosis with the correct one.

  • Conclusion: Using an LLM greatly simplifies development, as there’s no need to manually script all dialogue variations and symptoms.

2. Customer Interaction & Product Hypotheses

Current stage: The customer (Innopolis) has asked the team to independently formalize the requirements and propose product hypotheses that would be interesting to the market.

Dual audience: The team identified two key stakeholders:

  • The Customer Company (Innopolis): The source of real medical data and cases.
  • The University / Mentor: Evaluates the process and the application of the learned methodology.

  • Main risk: Signing legal documents directly with the customer could be dangerous. The contract might contain unfulfillable obligations for which the customer could hold the students personally liable.

  • Solution and recommendation: Documents should be signed only with the university. The university would then sign a separate agreement with the customer. This protects the team and creates a clear chain of responsibility.

4. Evaluation Criteria and Roles

  • Key criterion: The final grade for the project will come primarily from the university, not the external customer.

  • Process importance: The university needs to see that the team is applying the learned methodology, using proper tools, working with processes, and can argue their decisions.

  • Customer relationship: The relationship with the real customer is seen as a “training ground.” Their opinion is important but is not the sole or deciding factor for the final grade. In case of conflict or lack of information, the team must be able to explain their actions to the university.

5. Development Strategy

  • No blockers: The team can and should start development without waiting for official documents to be signed or real medical cases to be provided by the customer.

  • Synthetic data: For the prototype, the team can use synthetic cases generated with ChatGPT. This allows them to debug the system and demonstrate a working concept.

  • Action plan: Formalize requirements, propose product hypotheses, and simultaneously start writing code based on synthetic data.

6. Organizational Matters

  • Mentor meetings: Regular weekly meetings with the mentor are required for reporting.

  • Problem escalation: If the team encounters unsolvable problems (e.g., legal issues), the mentor is ready to escalate them to the program leadership.

Table of contents

This site uses Just the Docs, a documentation theme for Jekyll.