Unified data
Wearables, nutrition, habits, workouts, symptoms and medical reports share the same personal context.
BIOXS — AI-Powered Health & Fitness App
A mobile app that turns scattered health data into understandable analyses and personalised recommendations. A Knowledge Graph and Graph RAG provide traceable context, while document RAG, tool calling and multimodal models let the AI consult reports, use tools and interpret text and images.
End-to-end development across every architecture layer: mobile application, data, backend, deterministic analytics engine and artificial-intelligence integration.
The project
Steps and sleep usually live in one app, meals in another and laboratory results in isolated documents. BIOXS brings those sources into a single experience so people can understand how their health evolves and which factors are related.
Verifiable calculations
Trends, anomalies, correlations and scores are calculated by a deterministic engine, outside the LLM's control.
AI with real context
The assistant queries data, relationships and reports through tool-calling, clinical RAG and Graph RAG.
The mobile experience
The app is organised into areas covering everyday tracking, optimisation, prevention and chronic conditions.
Daily dashboard
Global score, metrics and evolution over time.
Optimisation
Activity, recovery, nutrition and training.
Prevention
Markers, trends and signals that require attention.
Nutrition
Voice, text or photo logging and nutritional analysis.
Medical reports
Biomarker extraction and clinical context through RAG.
Contextual assistant
Answers grounded in real data, tools and reports.
How it works
BIOXS separates calculations from language generation. Artificial intelligence explains and contextualises results that have already been calculated, validated and versioned.
01
Personal data
Metrics, meals, workouts, surveys and reports.
02
Health Analytics
Checks quality and calculates trends, anomalies, relationships and scores.
03
Knowledge Graph
Retains concepts and relationships with time, evidence, confidence and provenance.
04
AI Gateway
Retrieves relevant context, generates the explanation and controls the output.
05
BIOXS experience
Presents scores, recommendations, reports and conversational answers.
Deterministic engine
Produces reproducible snapshots from prepared data. The LLM neither invents figures nor decides the evidence level.
Knowledge Graph
Acts as longitudinal memory: every relationship retains its method, time window, effect, lag, confidence, status and provenance.
Contextual AI
Graph RAG retrieves the useful graph neighbourhood and clinical RAG supplies report fragments to explain results with personal context.
Product and communication
In addition to the app and its infrastructure, the project includes a public website explaining BIOXS capabilities, privacy, plans and value proposition.
Demonstration
The video shows the real BIOXS flow, from the daily dashboard and synchronisation to nutrition, medical reports, analytics and the contextual assistant.
Documentation
A technical overview of BIOXS as a project: its product, architecture, deterministic analytics, Knowledge Graph and artificial-intelligence integration.