Practical AI Using edge models to find sensitive data
6 snips
Jun 13, 2024 
Ramin Mohammadi, AI and ML lead at TauSight and adjunct professor at Northeastern, discusses the critical intersection of AI and privacy, particularly in handling sensitive health information. He highlights the alarming rise in healthcare data breaches and the challenges of managing unstructured data. Ramin explains how edge AI models can enhance the detection of protected health information, addressing limitations of traditional methods. He shares insights on the promise of smaller models and federated learning in bolstering data security within healthcare.
AI Snips
Chapters
Transcript
Episode notes
PHI Data Breaches
- PHI data breaches are a significant problem, affecting millions of Americans.
- Hackers target this valuable data for monetary gain, exploitation, and extortion.
Causes of PHI Breaches
- Hacking accounts for a majority of PHI data breaches.
- A smaller percentage stems from incidents like stolen laptops or phishing emails.
Challenges with Traditional Tools
- Healthcare companies rely on traditional tools with regex patterns for data security.
- These tools struggle to detect "dark PHI" residing within networks due to pattern limitations.