Independent project · Use CaseBiomedical Hardware · Database Connectivity · Clinical Platform & Patient Management

Patient Monitoring Bracelet

An end-to-end remote monitoring system combining an IoT wristband prototype with a telemedicine web platform.

IoTESP32C++FastAPISvelteKitPostgreSQLMQTTDocker

GitHub Thesis Dissertation (PDF)

Heart rate72bpm

Monitoring active Updated now

10.0 / 10Final GradeHonours Degree (Matrícula de Honor)

100 HzSampling RateMAX30102 PPG Red/IR

RAM BufferOffline ResilienceZero loss on Wi-Fi drops

End-to-EndEcosystemC++ · MQTT · FastAPI · SvelteKit

01 · Sensors

I2C (100Hz)

02 · Edge Core

Wi-Fi · MQTT

03 · Broker

Ingest REST

04 · Cloud Server

HTTPS / Nginx

05 · Telemedicine UI

Selected Architecture Node

ESP32 LilyGO TTGO T-Display

Edge Processing & Circular Buffer

C++FreeRTOSNTP Clock SyncTFT 1.14" DisplayRAM FIFO

Engineering Function

Dual-core 240MHz SoC. Runs signal filtering, handles TFT screen UI, synchronizes timestamps via NTP, and maintains an in-memory circular buffer for offline resilience.

Data Pipeline Role

Packages metrics into compact JSON payloads. If Wi-Fi is lost, buffers readings locally and flushes automatically upon reconnection.

ENGINEERING CASE STUDY

System Architecture & Implementation Story

The Clinical & Engineering Challenge

Passive home telemonitoring with guaranteed reliability

Demographic aging demands continuous, non-invasive home monitoring that enables clinicians to supervise patient vitals without requiring technical intervention from the elderly.

The core challenge was designing a complete end-to-end system: from custom biomedical sensor circuitry and embedded firmware up to a multi-tenant clinical web platform. The system needed to guarantee data integrity across unstable home Wi-Fi networks, detect falls in real-time, and provide clinicians with actionable historical trends.

Passive telemetry without elder burden
Multi-sensor fusion: Optical PPG + 6-Axis IMU
Automatic fallback during wireless dropouts

Hardware test bench on breadboard — Laboratory breadboard circuit integrating the ESP32 TTGO T-Display and operational I2C sensors.

Embedded Firmware & Edge Processing

Dual-core ESP32, optical PPG sampling & offline circular buffer

The wearable device is built on an ESP32 LilyGO TTGO T-Display SoC with an integrated 1.14" color TFT screen. The C++ firmware orchestrates two primary biomedical sensors over an I2C bus:

MAX30102 Optical PPG: Emits red and infrared wavelengths at 100 Hz through capillary tissue to extract raw photoplethysmogram waveforms, estimating pulse (BPM) and blood oxygen saturation (SpO₂).
MPU-6050 6-Axis IMU: Samples triaxial accelerometer and gyroscope vectors. An edge algorithm computes signal vector magnitude (SVM) to detect abrupt impact spikes followed by post-fall immobility.
Resilient Circular Buffer: If Wi-Fi connectivity drops, readings are preserved in a local RAM ring buffer with synchronized NTP timestamps, avoiding data loss until the broker becomes reachable again.

C++ firmware with modular sensor drivers
NTP epoch timestamping for analytical consistency
Local TFT visual status and battery feedback

Wristband prototype measuring heart rate on finger — Live optical photoplethysmography acquisition and on-screen metrics presentation on the wearable prototype.

IoT Messaging & Decoupled Broker Pipeline

Lightweight publish/subscribe model using Eclipse Mosquitto

To optimize battery life and minimize network overhead on the wearable, the system implements MQTT through an Eclipse Mosquitto broker:

Structured Topics: Clear namespace partitioning (patient/{id}/vitals, patient/{id}/alerts/fall, patient/{id}/events).
Compact JSON Payloads: Efficient data serialization containing biometric metrics, signal quality indicators, and NTP timestamps.
Quality of Service (QoS 1): Critical events like emergency button triggers and detected falls use confirmed delivery to ensure reception by medical ingest services.

Asynchronous event decoupling
Minimal byte transmission overhead
QoS guarantees for patient safety alerts

Telemedicine Web Platform & Clinical UX

SvelteKit frontend, async FastAPI backend & role-based workflows

The clinical web application was built with a modern containerized microservice architecture using Docker Compose:

Asynchronous FastAPI Core: REST API written in Python with SQLAlchemy and Pydantic validation, storing long-term clinical records in PostgreSQL and temporal sessions in Redis.
SvelteKit Clinical Portal: Highly responsive frontend providing three role-tailored dashboards:
- Doctor Portal: Patient cohort overview, pathology tagging, alarm resolution center, and daily study charts with interactive time-series powered by Chart.js.
- Patient Portal: Accessible self-monitoring view with intuitive health status indicators and secure doctor chat.
- Administrator Portal: System configuration and access management.