Why this page matters: Research shows most people remain uncomfortable with providers relying on AI in healthcare decisions. Trust increases with demonstrated performance, clinician presence, governance mechanisms, and representative data. Every section below maps directly to those evidence-based trust drivers.
Original research that grounds our approach
NAVSurgical is built on a foundation of peer-reviewed research by our team—spanning perceptual expertise in surgery, digital surgery definition, and robotic surgical education.
SAGES Digital Surgery Working Group; Ali JT, Yang G, Green CA, et al. Surg Endosc. 2024. PMID: 38180541.
Green CA, Lin J, O'Sullivan P, Huang E, Higgin R. Surg Endosc. 2023;37:571-579. PMID: 35579701.
Green CA, Lin J, Higgins R, O'Sullivan PS, Huang E. Am J Surg. 2022;224(3):908-913. PMID: 35637018.
Green CA, Chern H, Rogers S, Reilly L, O'Sullivan P. Annals of Surgery Open. 2021;2(3):e076. PMID: 37635816.
Green CA, Mahuron K, O'Sullivan P, Harris H. Academic Medicine. 2019;94(10):1532-1538. PMID: 30998574.
Green CA, Chu S, Huang E, Chern H, O'Sullivan P. American Journal of Surgery. June 2019. PMID: 31208624.
Green CA, Chern H, O'Sullivan PS. Journal of Robotic Surgery. Jan 2019.
Green CA, Abrahamson D, Chern H, O'Sullivan P. J Grad Med Educ. 2018;10(5):491-493. PMID: 30377478.
Media
Anantha V. MedTech Intelligence.
Nasdaq.
UCSF Department of Surgery. February 2018.
UCSF Department of Surgery. December 2017.
Safety by design.
Real-time intraoperative guidance carries real responsibility. These commitments define the boundaries of what NAVSurgical does and does not do.
Surgeon Authority
Automation Bias Mitigation
Cognitive Load & Alert Fatigue
What trains the system.
Our data sources, their constraints, and their limitations.
Real Operative Video
Thousands of de-identified surgical videos capturing on-screen behaviors, visual cues and decision signals that distinguish expert performance.
Synthetic Data
Generated from small, high-quality clinical datasets using validated synthesis techniques. Synthetic data extends training coverage without compromising patient privacy—and we state clearly where it's used and where it's not.
Building representative data over time.
Patient trust in medical AI correlates with the use of representative, diverse data. Here is where we are today and where we're headed.
Where we are now
Where we're headed
Designed for the surgeon's cognitive reality.
The operating room is not a dashboard. Cognitive load is already maximal. Our interface starts from that reality.
Design Principles
Information is filtered to the surgeon's current context. The system shows less, not more—presenting only signals relevant to the active procedural phase.
Critical signals are designed to enter awareness without demanding direct attention. The surgeon's primary visual focus stays on the operative field.
Risk signals use a continuous scale with calibrated confidence, not red/green binary states that force premature categorization of ambiguous situations.
If the system loses confidence or encounters novel scenarios, it reduces its visual footprint rather than escalating alerts.
Testing & Validation Approach
Interface prototypes are tested with practicing surgeons in simulated intraoperative conditions before any clinical deployment.
Systematic analysis of the surgeon's decision-making workflow during target procedures to identify when information helps versus when it distracts.
Usability findings drive interface changes before the next testing cycle. No feature ships without surgeon validation.
Regulatory-aware design.
We don't claim a specific regulatory pathway prematurely. Our design decisions are informed by current guidance, and we plan to engage regulators early as the product matures.
FDA CDS Guidance
The FDA's Clinical Decision Support guidance, finalized in January 2026, explicitly addresses automation bias in time-critical decision support. NAVSurgical's design philosophy was informed by this guidance from the outset.
Software & OTS Documentation
We build on established software infrastructure with appropriate documentation. The FDA provides specific guidance on off-the-shelf software used in medical devices, and our quality system reflects those expectations.
Staged Regulatory Strategy
Our three-stage product maturity model is designed around regulatory reality. We start in lower-risk territory (capture and insight), build evidence, and approach higher-risk functionality (guidance) only with the validation to support it.
Built for the hospital IT environment from day one.
Cybersecurity and data governance are not afterthought features for intraoperative systems. The FDA's February 2026 cybersecurity guidance for medical devices sets expectations we design to meet—not scramble to add later.
Data Encryption
End-to-end encryption in transit and at rest. No operative video leaves the hospital network without explicit institutional authorization and de-identification protocols.
Access Controls
Role-based access with principle of least privilege. Audit logging for every data access event. Integration with hospital identity management systems.
Network Architecture
Designed for hospital network segmentation requirements. Operates within existing security zones. No external dependencies required during intraoperative use.
Compliance Readiness
HIPAA-aligned data handling practices. SOC 2 Type II target. Penetration testing and vulnerability management as ongoing practice, not periodic audit.