Inclusive & Assistive Products
ISDN2001/2002: Second Year Design Project
Wristband & Vibration
Research Questions
-
Which output medium, whether an app or an external product, offers greater convenience for the mother and necessitates less attention while she engages in other tasks?
-
Which vibration pattern provides sufficient reassurance while minimizing feelings of panic and psychological distress during alerts?
Method
Structured interviews and vibration pattern tests with 10 mothers were conducted to evaluate alert effectiveness. Based on feedback, we developed a dual-alert system using an open-source C++ wristband platform. The wearable device was integrated with our FastAPI backend and connected to the AI analysis model. We programmed distinct vibration patterns - slow pulses (500ms interval) for positive alerts ("Laugh") and rapid pulses (100ms interval) for negative alerts ("Crying", "Unwell"). The system was designed to trigger these real-time vibration notifications upon receiving classified audio inputs from the AI model.
Results
Our testing revealed that both the wristband and companion app served distinct but complementary roles in caregiver convenience:
-
Wristband Superiority for Immediate Alerts
-
90% of mothers preferred vibrations over phone notifications for urgent alerts ("Crying"/"Unwell")
-
Enabled hands-free awareness during tasks like cooking or driving
-
40% faster response times compared to app-only notifications (2.1s vs 3.5s average)
-
-
App Necessity for Contextual Information
-
100% of users consulted the app after wristband alerts for details
-
Provided essential context (e.g., cry duration, predicted reason)
-
Served as a configuration interface for vibration sensitivity/patterns
-
User testing revealed a strong preference (90% approval) for differentiated vibration patterns, with mothers reporting reduced stress when distinguishing between positive and urgent alerts. The implemented system successfully delivered real-time notifications with 98% reliability in our stress tests. Positive alerts ("Laugh") were described as "reassuring" by 8/10 participants, while urgent alerts prompted immediate attention without causing panic. Technical validation showed the wristband maintained consistent connectivity with the FastAPI backend, with vibration triggers executing within 300ms of AI model classification. The solution effectively balanced emotional nuance with functional urgency in caregiver notifications.
Conclusion
Further research could explore:
-
Personalization of vibration patterns based on user preference
-
More complex features or alert system on wristband
In summary, this technical implementation advances infant monitoring systems by combining wearable technology, adaptive machine learning, and thoughtful haptic design—offering a scalable solution that prioritizes both safety and user experience.
This study successfully demonstrated the integration of a wearable wristband with a FastAPI backend and machine learning model to create an intelligent infant monitoring system capable of delivering context-aware haptic alerts. The technical implementation achieved three key objectives:
-
Seamless System Integration
-
Differentiated Haptic Feedback ("Laugh", "Crying," "Unwell")
-
Adaptive Machine Learning Coordination