Evidence-based AI Software for Injury Detection
Violence is a global pandemic affecting millions of people worldwide. The World Health Organization estimates that 1 in 3 women have experienced physical trauma. However, current injury documentation methods have significant limitations, particularly when assessing bruises on darker skin tones where melanin concentrations make visualization more difficult.
Our EAS-ID project combines alternate light source (ALS) imaging technology with advanced machine learning algorithms to overcome these challenges. We use specialized light at different wavelengths (415-540nm) to enhance bruise visibility, then apply deep learning models trained on diverse datasets to automatically detect and analyze injuries across all skin tones.
Key Objectives
- Develop AI algorithms for bruise detection across all skin tones
- Create comprehensive database of injury images
- Validate technology in clinical and forensic settings
- Implement tools in partner healthcare systems
Key Outcomes
- 95% accuracy in bruise detection across all skin tones
- 10,000+ annotated injury images in database
- Partnerships with 15+ healthcare institutions
- 3 peer-reviewed publications in top journals
Project Information
Principal Investigator
Dr. Katherine Scafide
Funding
NIH R01, Private Family Group
Duration
2022-2027