Why do you need an assistant?
Around Spina Bifida, families have many urgent and sensitive questions: what specialists may be needed, how to prepare for a consultation, where to find foundation materials, how to navigate rehabilitation, care, routing and social support.
A conventional generative model is not sufficient for such a task. The assistant must respond based on a previously prepared and verified knowledge base, foundation documents, reference materials and expert recommendations.
How does a RAG environment work?
Knowledge Base
Fund documents, reference materials and expert responses are structured into a managed database.
Search
The user's question is compared with relevant fragments of documents, and does not go into free improvisation.
Answer
The assistant generates a clear answer based on the sources found and can lead to fund materials.
Control
Risky medical formulations are limited and the knowledge base is updated through a controlled process.
Help scripts
For families
Explain basic concepts in simple language, help prepare for consultations and find foundation materials.
For patients
Quickly find reference information on care, routing, support, and available materials.
For doctors and experts
Provide unified access to a structured knowledge base and materials used in information support.
For NPOs
Reduce the burden on operators and consultants, maintain a consistent style of initial responses, and find internal materials faster.
Technologies
| Layer | What is used |
|---|---|
| RAG | Collecting documents, breaking them into semantic fragments, vector index, searching for relevant sources and generating a response based on materials. |
| Integrations | Foundation website, Telegram bot, CRM/Helpdesk if necessary, administrative panel for managing the knowledge base. |
| Quality | Limiting risky medical statements, checking the knowledge base, controlling sources and the ability to retrain scenarios. |
| Safety | Minimization of personal data, differentiation of access rights, logging, deleting data upon request and working with sensitive information in accordance with the requirements of Federal Law No. 152-FZ. |
Medical disclaimer
The chatbot is not medical advice, does not prescribe treatment, and does not replace seeing a doctor. It provides reference information based on a knowledge base and helps you quickly find relevant materials from the foundation and expert community.
Platform framework RESTART AI Platform
The RAG assistant was developed using the RESTART AI Enterprise Platform: the RESTART platform AI environment, which allows you to collect verified sources, manage the knowledge base, limit risky language, keep logs and connect convenient interaction channels.
The platform approach helps not to reduce the project to one chatbot: an outline of data, rules, sources, roles, updating knowledge and quality control of answers appears around the assistant. The same approach underlies 1trAIner and other RESTART applied AI products.
AI Compute for RAG assistants
The foundation's RAG assistant shows that a knowledge base needs infrastructure as much as good text: document loading, vector index, updates, logs, risky wording restrictions, roles and backup must be part of the solution.
Partnership
RESTART is open to partnerships with healthcare organizations, foundations, research groups, regional teams and technology partners who are developing solutions for early intervention, routing and support for families faced with a diagnosis of Spina Bifida.
Let's discuss your environment
Describe the task, current systems, constraints, and expected results. We will offer a practical first step: diagnostics, pilot, audit, roadmap or project team.
