Symbiocare

SymbioCare: Human-AI Hybrid Support for Dementia Care

SymbioCare is a cutting-edge system designed to revolutionize dementia care by combining the efficiency of AI with the empathy and oversight of human caregivers. Utilizing a Human-in-the-Loop Retrieval-Augmented Generation (RAG) model, SymbioCare ensures critical patient care decisions are always validated by human experts before execution.

Overview

SymbioCare is designed to provide personalized, responsive care for dementia patients, ensuring safety and well-being while empowering caregivers to collaborate with AI. The system continuously monitors patients' health data, routines, and behaviors via wearables, home sensors, and medical records. Using a Human-in-the-Loop (HITL) approach, the AI suggests actions, which are reviewed and approved by caregivers before being executed.

Features

1. Smart Alerts & Assistance

• Continuous monitoring of patients using wearables and home sensors.
• Detects critical behaviors such as wandering, medication non-compliance, or unusual activity patterns.
• Notifies caregivers for approval before executing actions like sending alerts or administering medication reminders.

2. Personalized AI Caregiver

• AI adapts to the patient's specific routines, behaviors, and preferences.
• Offers personalized recommendations, such as calming activities during agitation or reminders based on health conditions.
• Human caregivers can review, approve, or adjust AI-generated recommendations.

3. Dynamic Human-AI Collaboration Dashboard

• Caregivers oversee AI’s decisions through an intuitive dashboard.
• Human experts validate or modify AI actions before execution.
• Caregivers and AI work together to ensure personalized and adaptive care.

4. Family & Caregiver Engagement

• Real-time notifications and health reports for family members and caregivers.
• Human caregivers approve any major decisions to ensure safe and empathetic patient care.

5. Behavioral and Emotional Insights

• AI leverages speech and emotion recognition to detect patient mood changes or anxiety episodes.
• Caregivers are prompted to intervene when emotional distress is detected.

6. Localization and Cultural Sensitivity

• Localized care techniques that are sensitive to cultural preferences and individual patient needs.
• Human caregivers help refine these personalized care strategies.

System Architecture

SymbioCare is built using the Flowise framework to incorporate Human-in-the-Loop (HITL) RAG models. The system combines multiple AI tools to process patient data, generate recommendations, and ensure human approval before taking action.

High-level components:

1. OpenAI Embeddings - Generates vector embeddings for patient data.
2. Pinecone - Vector database for storing and retrieving historical patient information.
3. ChatOpenAI - Used for generating personalized care recommendations based on retrieved information.
4. Agent Memory - Stores and retrieves data related to patient history and caregiver actions.
5. Notification API - Sends alerts and notifications once actions are approved.
6. Human Approval Node (Agent) - Ensures that all critical AI-generated actions are approved by caregivers before execution.

Technologies Used

• Flowise: Visual tool for building human-in-the-loop AI workflows.
• OpenAI: For embedding and generating personalized care suggestions.
• Pinecone: Vector database for fast, scalable memory storage.
• ChatOpenAI: Chat-based interface for generating recommendations.
• Agent Memory: Stores historical data for personalized care.
• Notification API: Sends alerts and notifications to caregivers and family members.

How It Works

1. Data Collection:

   • The system collects real-time data from patient wearables, home sensors, and health records.

2. AI Suggestions:

   • Using the RAG model, AI retrieves relevant information from memory and generates personalized recommendations (e.g., medication reminders or calming activities).

3. Human-in-the-Loop Approval:

   • Before any action is executed, the system sends the AI-generated suggestion to the human caregiver for review.
   • The caregiver approves, rejects, or modifies the suggestion through the system’s dashboard.

4. Action Execution:

   • Upon approval, the system executes the suggested action (e.g., sending a medication reminder or alert).

5. Continuous Learning:

   • The system continually learns from human feedback, improving future suggestions and actions based on evolving patient needs.

SymbioCare empowers caregivers to deliver the highest level of personalized care, ensuring that AI decisions are always safe, empathetic, and human-approved.