"Prescription is Science. Administration is Human."
Dia-Smart is a dual-node IoT ecosystem designed to bridge the gap between medical prescription and patient compliance. It addresses the "Last Mile" problem in diabetes care by automating the tracking of insulin dosage, storage conditions, and blood glucose trends for elderly and Type 1 diabetic patients.
Unlike standard logbooks, Dia-Smart creates a closed-loop system that correlates prescribed doses with actual injected units and physical inventory.
Diabetes management requires precise coordination between blood sugar testing and insulin administration. However, current solutions have major "blind spots":
- [cite_start]Dosage Errors: Elderly patients often forget if they took their shot or dial the wrong dose due to poor vision[cite: 8, 10].
- [cite_start]Manual Logging: Patients often fail to accurately record their readings in logbooks[cite: 26, 28].
- [cite_start]Unsafe Storage: Insulin kept in domestic fridges is often subject to spoilage without the user knowing[cite: 20, 21].
Dia-Smart separates the Storage from the User Interface, creating a distributed system that monitors the patient at every step:
A retrofit smart cap that clips onto standard insulin pens to automatically log injection data.
-
Inertial Sensing: Uses an MPU6050 Gyroscope to detect the exact angle of dosage dialing (
$15^\circ \approx 1 \text{ Unit}$ ), distinguishing between dialing up and correcting down. - Injection Detection: Accelerometer-based gesture recognition detects the specific motion of injection.
A retrofit module that upgrades non-smart glucometers (specifically Mega Check TD-4257) into IoT devices.
- Protocol Hacking: Intercepts RS-232 serial data from the glucometer's stereo jack to read historical values.
- [cite_start]Auto-Sync: Automatically fetches glucose readings + timestamps and uploads them to the cloud via Wi-Fi[cite: 58].
A passive sensing enclosure placed inside the refrigerator.
- [cite_start]Inventory Logic: High-precision Load Cells detect if the insulin pen is present and estimate remaining cartridge levels[cite: 62].
- [cite_start]Safety Monitoring: DS18B20 sensor alerts guardians if the fridge temperature deviates from the safe range (
$2^\circ\text{C} - 8^\circ\text{C}$ )[cite: 60].
- [cite_start]Real-time Dashboard: Visualizes blood sugar vs. insulin dosage graphs[cite: 64].
- Guardian Alerts: Push notifications for missed doses, hypoglycemia events, or critical inventory levels.
| Domain | Technology / Component |
|---|---|
| Microcontrollers | ESP32-C3 SuperMini (Wearable), ESP32 DevKit V1 (Base) |
| Sensors | MPU6050 (IMU), HX711 + Load Cell, DS18B20 (Temp), MAX3232 (RS232-TTL) |
| Connectivity | BLE 5.0, ESP-NOW, MQTT over Wi-Fi |
| Mobile App | Flutter (Dart) |
| Backend | Firebase (Firestore, Cloud Functions) |
The system operates on a Master-Slave topology:
- The Edge (Smart Cap): Sleeps 99% of the time. Wakes on movement to record dosage.
- The Hub (Base Station): Always powered. Connects to the Glucometer and acts as the Wi-Fi gateway for the Cap.
- The Cloud: Processes data and triggers alerts to the Guardian's phone.
- E/21/031 - Ananthasagaran N. (Email)
- E/21/036 - Arnikan U. (Email)
- E/21/356 - Sanjeevan U. (Email)
- E/21/386 - Sivasuthan J. (Email)
Supervisors:
- Dr. [Supervisor Name]
- Dr. [Co-Supervisor Name]
This project is licensed under the MIT License - see the LICENSE file for details.