Vital AI:
AI-Integrated IoT Health Monitoring
for Enhanced Elderly Care
By:-
Chiranth S
Namratha V J
Nisha S
Shravan M R
preencoded.png
Introduction
As the elderly population grows, there is an urgent need for accessible, real-time
health monitoring solutions. This project leverages a combination of IoT sensors
and Artificial Intelligence (AI) to provide continuous monitoring of vital health
parameters like heart rate, SpO₂, temperature, and fall detection.
Our system not only collects data using low-cost sensors and transmits it to the
cloud via ESP32 and Firebase, but also integrates AI algorithms to detect
anomalies, identify patterns, and trigger timely alerts. This fusion of IoT and AI
ensures proactive care, personalized health insights, and improved emergency
response for vulnerable individuals.
preencoded.png
Problem Statement
Elderly individuals often lack access to real-time health
monitoring, as traditional systems tend to be costly, lack
personalization, and depend heavily on manual intervention.
Critical incidents such as falls or sudden health anomalies
frequently go unnoticed, delaying necessary response. This
highlights the urgent need for an affordable, AI-integrated IoT
solution that enables continuous monitoring, proactive care,
and timely emergency alerts.
preencoded.png
Objectives
• Provide continuous health monitoring for elderly and
physically challenged individuals.
• Enable real-time detection of anomalies (e.g., falls,
abnormal vitals).
• Deliver a low-cost, customizable, and user-specific
solution.
• Integrate AI for proactive alerts and personalized insights.
• Contribute to healthcare innovation with real-world and
research impact.
preencoded.png
Future enhancement
•Planned integration of ML for anomaly detection and event
prediction.
•Use of LLMs for contextual decision-making and care
recommendations.
•Image processing for object/fall detection using ESP32-CAM.
•Audio event detection for keywords like “Help” or sound-based
diagnostics.
preencoded.png
Development Process
1. Sensor & Data Collection Layer:
- Collects data from sensors.
- ESP32 sends structured real-time data to Firebase
2. Agent Architecture & AI Integration:
- Memory: Firebase stores logs
- Planning: Predictive alerts for health issues
- Future: LLMs for multi-modal pattern interpretation
3. Virtual Interface & Visualization:
- Dashboard for live monitoring and alerts
- Caregiver interaction with logs and settings
preencoded.png
Development Process
4. Integration & Deployment Pipeline:
- Real-time connection between ESP32, Firebase, and UI
- Cloud functions for real-time alerts
- Event-driven: fall, ECG spike, cough, etc.
5. Advanced Features (Planned Enhancements):
- Audio Detection: Distress signals, cough detection
- Image Recognition via ESP32-CAM
- Predictive Analytics using historical data
- LLM-Driven Reasoning and Decision Support
preencoded.png
System Architecture
Microcontrollers Health Sensors
• ESP32: Dual-core, Wi-Fi, • MAX30100: SpO2, HR
Bluetooth 4.2 accuracy ±3%
• Arduino UNO: • BPM108: Single-lead ECG,
ATmega328P, 16 MHz 500 Hz sample rate
clock • ADXL345: 3-axis
accelerometer, ±16g
Location
GPS NEO-6M module with 2.5m accuracy
preencoded.png
Current Implementation
➢ Hardware Integrated
➢ Real-time data from all sensors
➢ Firebase integration for cloud storage
➢ Live ECG graph via Arduino UNO
➢ Basic mobile app for caregiver access
Unique Features
➢Low-Cost Sensors – Avoids expensive devices like Apple Watch
➢AI + Historical Data – Personal health baselines, not generic
➢Remote Access – Caregivers can view live metrics
➢GPS + Alerts – Falls or abnormal vitals trigger location alerts
preencoded.png
preencoded.png
