System
Source (fictional): PDF or note text describing a 41-year-old with fibroids, heavy
bleeding, and anemia who ultimately undergoes minimally invasive hysterectomy.
De-ID AI
In the live system, this module would:
1. Strip names, MRNs, addresses, contact info, and direct identifiers.
2. Normalize age and dates as needed (e.g., “early 40s,” “late 2024”).
3. Convert the clinical course into a structured, case-list ready format with
clearly labeled indications, findings, procedure steps, and outcomes.
4. Emit a JSON or table representation ready for training and analytics.
Mock transformation summary · Not real de-identification logic
De-ID AI
Example of a downstream representation (fictional):
• Age band: 40–44
• Indication: Symptomatic leiomyomas with AUB-L and iron-deficiency anemia
• Procedure: Total laparoscopic hysterectomy with bilateral salpingectomy
• Key risk factors: BMI > 35, prior C-section ×1
• Outcome: Same-day discharge, EBL < 100 mL, benign pathology
This is the type of clean, reusable case that can safely train future PDI models.
Fictional example · Safe for demos
