Falls are the #1 cause of injuries in older adults. Falls can be due to external perturbations to balance, such as slips or trips, or internal causes such as incorrect weight shifting, or loss of support with an external object. Understanding the cause of the fall is essential in guiding approaches to prevention. This project will help in providing the scientific basis for 'smart cameras' that can detect the occurrence of a fall, and provide care providers with information on how and why the fall occurred. Using an extensive library of video footage of real-life falls which has been manually encoded (for outcomes including cause of fall), we will examine the accuracy of video language models (e.g., GPT and Gemini) in identifying the cause of falls, the activity at the time of the fall (e.g., transferring, walking, or standing), and the consequences of the fall in terms of diagnosed injuries.
Required Qualifications:
Candidates should be passionate about learning and immersing themselves in the research environment. They should have a strong scientific curiosity, and possess skills in data analysis and oral and written communication. They should be detail-oriented, careful in their approach to research, and able to work well both independently and in a collaborative environment.
Recommended Qualifications:
Previous research experience is a plus. Strong quantitative skills in fields including computer programming and biomechanical analysis.
Semester: Summer 2026
