MIT EECS — Accenture Undergraduate Research and Innovation Scholar
Using Data Analytics to Improve Wheelchair User Autonomy and Safety
Dorothy W. Curtis
Why can’t a wheelchair be both comfortable and intelligent? This was the rationale in integrating an unobtrusive sensor system with a wheelchair. This system has been deployed at The Boston Home, resulting in 3 Gigabytes of data for further analysis and algorithm development. My goal is to make the wheelchair truly intelligent by analyzing real-time data with signal processing and machine learning to generate useful alerts for users and caregivers. Applications include preventing user fatigue, heat exhaustion detection, pressure sore alleviation and recognition of sleep apnea through analysis of respiration rate, heart activity and external conditions. This gives users more freedom and information, while providing the safety net of a caregiver alert system.
My previous UROP in the Lodish lab used web scraping to identify metabolic proteins. At Johnson & Johnson, I designed and implemented visualization tools for complex drug production systems. Most recently, at Smarsh, I used machine learning to analyze large datasets for fraud. I hope to compile these experiences in machine learning, analytics and visualization to create an interface between user, caregiver and wheelchair.