NeuroSense is an Arduino embedded firmware repo with two wearable-style prototype modules:

• An ESP8266 biometric monitor that reads heart rate, SpO2, and temperature, then publishes telemetry to Azure IoT Hub and ThingSpeak.
• An ESP32 sleep-stage alarm that reads heart rate and motion, classifies sleep state with a heuristic rolling-window classifier, streams output over Bluetooth Serial, and drives a vibration motor during light-sleep wake windows.

This project is for education, prototyping, and engineering experimentation only. It is not for medical use, diagnosis, treatment, safety-critical monitoring, or clinical sleep staging.

MAX30100 / MLX90614 / MPU6050
    -> ESP32 / ESP8266 firmware
    -> filtering + heuristic sleep classifier
    -> MQTT / HTTP / Bluetooth Serial
    -> Azure IoT Hub / ThingSpeak / terminal review

The biometric firmware uses placeholder-safe configuration, sensor-ready flags, bounded Wi-Fi/MQTT retry behavior, and validity-tagged Azure telemetry. The sleep alarm firmware uses guarded sensor handling and a 30-second heuristic classifier window. Production TLS certificate validation is still a TODO; the ESP8266 Azure client currently uses development-only TLS mode.

C:\Neurosense
|   README.md
|   platformio.ini
|   .gitignore
|
+---Biometric Monitor
|   |   README.md
|   |   azure_thingspeak_setup.md
|   |
|   +---Biometric_Monitor
|   |       Biometric_Monitor.ino
|   |
|   \---lib
|       \---MAX30100_PulseOximeter
|           \---Arduino-MAX30100-master
|
+---Sleep Stage Alarm
|   |   README.md
|   |
|   \---sleepstage_wake
|           sleepstage_wake.ino
|
+---include
|       secrets.h.example
|       sleep_classifier.h
|
+---src
|       biometric_monitor.cpp
|       sleep_stage_alarm.cpp
|       sleep_classifier.cpp
|
+---tests
|       test_sleep_classifier.cpp
|
\---docs
        build_notes.md
        phase_summary.md
        security_notes.md

• Main file: Biometric Monitor/Biometric_Monitor/Biometric_Monitor.ino
• Board: ESP8266 NodeMCU
• Sensors: MAX30100 and MLX90614
• Outputs: Azure IoT Hub MQTT telemetry and ThingSpeak HTTP updates
• Reliability: bounded Wi-Fi setup/retry, MQTT backoff, sensor-ready flags
• Telemetry: BPM, SpO2, object temperature, ambient temperature, metadata, status code, sensor-ready flags, and validity flags

• Main file: Sleep Stage Alarm/sleepstage_wake/sleepstage_wake.ino
• Board: ESP32 Dev Module
• Sensors: MAX30100 and MPU6050
• Outputs: Bluetooth Serial diagnostic stream
• Actuator: vibration motor on GPIO25 through a transistor driver
• Classifier: heuristic rolling-window classifier over about 30 seconds
• Tests: tests/test_sleep_classifier.cpp contains standalone host-readable classifier vectors

From the repo root:

pio run -e esp8266_biometric_monitor
pio run -e esp32_sleep_stage_alarm

The PlatformIO environments are defined in platformio.ini. They use wrapper files in src/ so the original Arduino .ino files stay in place.

The original .ino sketches remain in their Arduino-friendly folder layout:

• Biometric Monitor/Biometric_Monitor/Biometric_Monitor.ino
• Sleep Stage Alarm/sleepstage_wake/sleepstage_wake.ino

For Arduino IDE builds, install the board package and libraries listed in each module README. If the IDE does not see the repo-level include/ folder, copy your local secrets file into the biometric sketch folder as secrets.h.

1. Copy include/secrets.h.example to include/secrets.h.
2. Fill in local Wi-Fi, Azure IoT Hub, SAS token, ThingSpeak, and firmware version values.
3. Never commit include/secrets.h.

include/secrets.h is ignored by Git. Azure SAS tokens expire and should be regenerated as needed.

• PlatformIO build environments for ESP8266 and ESP32.
• Placeholder-based secret handling via include/secrets.h.example.
• Biometric metadata and validity-tagged Azure payloads.
• ThingSpeak fields for biometric readings plus validity/status fields.
• Bounded Wi-Fi setup/retry and MQTT retry backoff in the biometric firmware.
• Sensor-ready flags and non-locking sensor failure handling.
• ESP32 heuristic sleep classifier using a 30-second rolling sample window.
• Standalone classifier test vectors.

• Replace ESP8266 setInsecure() with CA certificate validation before production use.
• Add a configured PlatformIO test target for the sleep classifier.
• Tune sleep classifier thresholds using real labeled data.
• Reduce noisy Serial output if needed during longer sessions.
• Add optional offline buffering only after defining storage limits and failure behavior.

• Not medical-grade and not clinically validated.
• Sleep states are heuristic estimates, not diagnostic sleep stages.
• Azure TLS certificate validation is not yet implemented.
• No battery voltage telemetry is implemented because no battery sensing circuit is defined in firmware.
• No mobile app, dashboard screenshots, or cloud analytics are included in this repo.