From the AWS edition of November 9, 2026
AWS profiles Xtasis, a Northwestern researcher's company, whose de-identification and citation tools run on AWS, with claimed 60-75% literature-time savings
Amazon Web Services (profile of Xtasis / Northwestern University researcher) · Higher education research enterprise / medical research · United States (Northwestern University, Illinois) with users worldwide (claimed)
- Publisher
- AWS Public Sector Blog
- Original publication
- 2026-11-09
- Source retrieved
- 2026-11-10
- Event date
- 2026-11-09
What happened
AWS's November 9 Public Sector Blog post profiles three tools built by Xtasis, the company of Northwestern research assistant professor Adrienne Kline. PixelGuard de-identifies medical images, with a crosswalk log and low-confidence changes flagged for human review, and deploys inside the customer's own environment; NoteGuard redacts all 18 HIPAA identifiers from clinical notes. Both are sold through AWS Marketplace. CiteGeist manages literature, verifies that references exist and uses NVIDIA Nemotron via Bedrock and SageMaker AI for summaries. Claims include 'hundreds of hours' of redaction saved and a roughly 60-75% cut in literature-management time; no method or sample is given.
Why it matters
Relevant to research universities and academic medical centers preparing clinical data for NIH-funded studies and to research-integrity offices concerned about fabricated citations. Cross-tagged to the research stream because the use case is the university research enterprise. Northwestern is private; applicability to public universities is plausible but not shown.
Evidence and measured results
Vendor customer story with company quotes; metrics are self-reported without methodology, baseline or independent validation.
Limitations and uncertainty
Vendor marketing with unverified metrics; private-university origin; no de-identification accuracy figures.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-10; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Research teams spend months de-identifying clinical images and notes and managing citations, delaying studies.
- Stakeholders
- Vice president for research
- IRB / privacy office
- Research computing director
- Research integrity office
- Discovery questions
- How is de-identification done and validated today?
- Is there an AWS account approved for PHI with a BAA?
- How are citation errors detected now?
- Value hypothesis
- In-account de-identification and citation tools may shorten study preparation; savings are self-reported.
- Potential engagement
- Validation pilot on a labeled sample of clinical images and notes.
- Unsupported claims
- 60-75% time savings as an established result
- De-identification accuracy
- Applicability to public universities
Pre-sales engineering
Role takeaway
- Fit
- Good architectural fit for universities with AWS research accounts.
- Architecture and integration
- Marketplace deployment into the institution's account using Cognito SSO, ECS Fargate and KMS; CiteGeist uses Bedrock and SageMaker AI.
- Prerequisites
- PHI-approved AWS account
- BAA
- KMS key policy
- SSO integration
- Constraints
- No published accuracy
- Private-university origin
- Pricing not stated
- Security
- PHI stays in the customer account; validate redaction completeness and protect crosswalk logs.
- Proof of value
- Measure missed identifiers on labeled sample
- Measure false redactions
- Compare reviewer time with current process
Delivery
Role takeaway
- Work
- Run a de-identification validation pilot with privacy oversight.
- Dependencies
- IRB approval
- Labeled sample
- Reviewers
- Ownership
- Research computing, with privacy office oversight.
- Skills and adoption
- Reviewers need training on crosswalk logs and low-confidence flags.
- Governance checkpoints
- Accuracy review
- Crosswalk-log retention rules
- Approval before production cohorts
- Acceptance criteria
- Proposed: identifier miss rate at or below agreed threshold
- Proposed: reviewer time measured against baseline
- Risks
- Re-identification from missed identifiers
- Over-reliance on AI summaries
- Unverified vendor metrics
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
PixelGuard: CloudFront/S3 front end, Cognito, API Gateway, Lambda, DynamoDB, EventBridge, ECS on Fargate, KMS, WAF, CloudWatch; deployed in the customer's own account. CiteGeist: Amplify, API Gateway, Lambda, Step Functions, Nemotron via Bedrock and SageMaker AI.
Governance
Who approves, reviews and stays accountable for outcomes?
In-account deployment keeps PHI under institutional control; IRB and privacy offices still need to validate de-identification quality.
Security and privacy
What data, permissions and controls need testing?
Customer-account deployment reduces data sharing; residual re-identification risk from missed identifiers is not quantified.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Not addressed.
Procurement
What should contracts, pricing and exit terms secure?
PixelGuard and NoteGuard are available in AWS Marketplace; pricing not stated.
Operating model
Which teams own the service once it runs?
Human review of flagged changes remains part of the workflow.
Publication history
- 2026-11-09AWS · Issue 118 resources
Stable resource ID: aws-xtasis-northwestern-research-deidentification-citegeist-20261109