P
Privacy Policy & Data Protection
PJKPredict: Spine Risk Radar • Effective Date: September 2026
Privacy Commitment: PJKPredict is engineered by an academic neurosurgeon and spine surgeon with strict medical data privacy, institutional HIPAA compliance principles, and physician-device containment as core design pillars.
1. Protected Health Information (PHI) & HIPAA Compliance
HIPAA Safe Harbor De-Identification
PJKPredict operates strictly as a physician clinical decision-support and surgical planning research instrument. In alignment with HIPAA Safe Harbor De-Identification standards (45 CFR § 164.514(b)):
- Automatic Metadata Stripping: When an imaging study is selected via camera or photo library, all embedded file metadata—including DICOM tags, EXIF records, GPS coordinates, device serial numbers, and patient demographic headers—is automatically excised and stripped on the physical device prior to any processing.
- Visual De-Identification: Clinicians are instructed to utilize radiographs without burned-in visual identifiers (patient name, MRN, date of birth). The application only analyzes radiographic bony anatomical contours.
- Zero Patient Registry: The app does not collect, record, index, or store patient names, medical record numbers, hospital accounts, or demographic identities.
2. AI Radiographic Landmark Processing Architecture
PJKPredict offers optional multimodal artificial intelligence assistance for automated spinopelvic landmark detection (e.g., C7 plumbline, S1 promontory, bicoxofemoral axis):
- Ephemeral Edge Execution: Radiographic inference requests are routed through a secure, serverless edge isolate over TLS 1.3 encryption.
- Zero Server-Side Storage: The edge proxy operates entirely in volatile memory. No images, landmarks, prompts, or results are written to disk, databases, object stores (S3/R2), or persistent caches.
- Immediate Discard: Once landmark coordinates (normalized percentage points) are computed, the image payload is instantly purged from memory.
- Zero AI Model Training: Under developer enterprise API agreements, customer payloads are never used to train, retrain, or improve artificial intelligence or computer vision foundation models.
3. On-Device Biomechanical Simulation
All core clinical algorithms—including the Yilgor Global Alignment and Proportion (GAP) Score, Lafage-ISSG logistic risk models, Schwab age-adjusted alignment corridors, and CT Hounsfield Unit bone mineral density calibrations—execute 100% locally on your iOS device using native Swift code without requiring internet access or cloud communication.
4. Analytics, Tracking & Third Parties
- No Commercial Tracking: PJKPredict contains zero third-party tracking frameworks, zero advertising SDKs, and zero telemetry monitors.
- No Data Brokering: We do not sell, rent, license, or monetize physician or clinical data under any circumstances.
- No Account Required: The application functions immediately without requiring user registration, email sign-in, or profile creation.
5. Physician Support & Contact
For questions regarding clinical data protection, regulatory documentation, or technical security architecture, please contact:
A. Alex Mohit, MD, PhD
Practicing Complex Spine Surgeon & Academic Neurosurgeon
✉️ Email: mohalex@gmail.com
Phone: +1 (206) 679-5710