Connect AI assistants to your data
Let approved users ask questions of De-Identified EHR data through compatible AI chat tools, without exposing PHI or waiting for governance to approve it.
September 14–17 · Hilton Chicago
Find us at booth 5122Meet Subsalt at Becker’sHealth IT + RCM Conference
See AI work with patient data—without accessing PHI—at booth 5122.
Try a live AI chat demo using a De-Identified Digital Twin. Explore what this could mean for your health system’s AI initiatives, while source PHI stays inside your environment.
Request a meeting
30 minutes · Booth 5122
We’ll confirm your time by email.
Meeting request received.
Our team will email you to confirm a time. Your meeting is not booked until you receive that confirmation. In the meantime, the whitepaper explores healthcare data access and AI.
That didn't go through.
Your request hasn’t been sent. Your answers are still in the form. Please try again, or email hello@getsubsalt.com and we'll set it up by hand.
The conferenceBecker’s Health IT + RCM
When & where
September 14–17, 2026 · Hilton Chicago
Meet the Subsalt team
Start with your priority
Advance your health
system’s AI initiatives.
Expand what your teams can do with health data, without giving them direct access to source PHI.
Request a meetingTrain AI models for your health system
Use De-Identified training data derived from your structured EHR data to develop internal models for clinical and revenue-operations use cases.
Evaluate analyses against source data
Run Python in Secure Runtime to calculate results from source records and receive De-Identified analyses, without direct access to PHI.
Develop data assets
License De-Identified data derived from your own records, on your terms. The engine runs inside your environment, so the arrangement stays between your health system and the licensee, with no data broker in the middle.
One privacy-preserving platform
Expand data access.
Keep source PHI inside.
One Subsalt deployment connects a queryable digital twin with secure analysis against source data. Both run inside your health system’s environment.
Your health system’s environmentOne Subsalt deployment · Source PHI stays inside
Source PHI (EHR Data)
Structured records stay inside your environment.
Source data supports two paths below
Subsalt · One deployment
01 · De-Identification Engine
De-Identified Digital Twin
Expert Determination by Datavant. Researchers and AI query the twin through SQL or compatible AI chat.
Engine → Digital Twin → SQL / AI chat
02 · Secure Runtime
Python analysis on source PHI
Python runs against source data inside Subsalt. Outputs are checked for de-identification before users receive results.
Source PHI → Runtime → De-Identified charts
SubsaltDe-Identification Engine + Secure Runtime
Source PHI (EHR Data)Structured records · Rows & columns
De-Identification EngineConfigured for your data
De-Identified Digital TwinExpert determination by Datavant
SQL / AI chatAccess for researchers & AI
Secure RuntimePython analysis on source PHI
De-Identified ResultsCharts, heatmaps & scatter plots
Explore through SQL or compatible AI chat. Analyze source data with Python. Discuss the right approach for your initiative.
Request a meetingBefore we meet.
What can I see at Subsalt’s booth?
See a live AI chat demo querying a De-Identified Digital Twin at booth 5122. Meet with our team to discuss your AI initiative and the data it needs. The booth demo does not imply general availability for every AI chat integration.
Is the digital twin synthetic data?
Yes. Subsalt generates the digital twin from patterns in your structured EHR data. It is designed to preserve relationships that make the data useful for exploration, research, and AI, rather than reproduce individual patient records. Synthetic describes how data is created; it does not, by itself, establish that the data is De-Identified.
Is Subsalt’s digital twin certified De-Identified?
Yes. Subsalt’s digital twin is De-Identified through Expert Determination by Datavant. Datavant independently evaluates Subsalt’s technology, and Subsalt configures the De-Identification Engine for your dataset. De-identification documentation accompanies outputs covered by the determination. The determination’s scope and conditions matter; certification is not a blanket approval for every possible use or combination with other data.
Is the data De-Identified under Safe Harbor?
Subsalt uses Expert Determination, the other de-identification method recognized by HIPAA. Safe Harbor requires removing 18 categories of identifiers and having no actual knowledge that the remaining information could identify a person. Expert Determination uses a qualified expert’s documented analysis to establish that identification risk is very small. It can support retaining detail that Safe Harbor would otherwise require removing, subject to the expert’s assessment.
Read HHS guidance on de-identification ↗
Does the digital twin contain the same patient records as the source?
The twin has a familiar table structure for querying structured EHR data, but it is not an exact copy of each patient’s record. Use it to explore cohorts and develop analyses. When an analysis needs calculations against source data, Secure Runtime runs Python inside Subsalt and checks outputs for de-identification before returning results.
Where is Subsalt at Becker's Health IT + RCM 2026?
Subsalt is exhibiting at Becker's 11th Annual Health IT + RCM Conference at the Hilton Chicago, September 14–17, 2026, at booth 5122. Meet the Subsalt team Monday through Wednesday, with additional availability Thursday morning.
How do I book a meeting with Subsalt at Becker's?
Send a meeting request using the form on this page. Our team will email you to confirm a time. Your request does not automatically reserve a calendar slot.
What does Subsalt do?
Subsalt runs inside your health system’s environment. It gives your institution a de-identification engine configured for your dataset, so teams can query a de-identified digital twin for AI, analytics, and research. Source PHI stays inside your environment.
Meet Subsalt at Becker’s
Bring your AI initiative.
See AI chat in action and discuss what your technical and governance teams should evaluate next.
Request a meeting
Not attending? Talk with our team ↗