Engine · Tidegate
The Tidegate engine

Find, protect and link health data where it lives.

One foundation and four layers. It shows you what your data holds, protects it and labels every field, without the data leaving your environment. The same labels carry forward into linkage.

engine run · 4 sources {{ heroClock }}
{{ h.title }} {{ h.out }}
TDG-4Q7X: found, protected, linked and labeled.
Illustrative. Fictional patient.
How it works

One foundation, four layers.

  1. {{ l.n }}

    {{ l.k }}

    {{ l.v }}

    Technical details
    • {{ p }}
      •
    IN{{ l.in }}
    OUT{{ l.out }}
    2.
Coverage

What the Tidegate engine recognizes

Sensitive health data rarely arrives labeled. The Tidegate engine identifies it across structured records, unstructured clinical notes and embedded documents, then classifies every finding by what it is, how sensitive it is and which rules are mapped to it.

DetectedOn the roadmap
01

Direct and indirect identifiers

Who a record is about, where, when and how to reach them.

PEOPLE
Names, with patients and clinicians told apartUsernamesProfessions
PLACE
Street addressesCitiesCountiesZIP codesStatesCountriesGPS coordinatesFacilitiesUnitsOrganizations
TIME
Dates of birthAdmission datesDischarge datesService datesDeath datesAgesTimesDurations
CONTACT
Phone and fax numbersEmail addressesURLsIP and device addresses
ID NUMBERS
Social Security numbersMedical record numbersMember numbersClaim numbersCase numbersAccount numbersLicense numbersPassportsState IDsVehicle IDsDevice IDsAccession numbers
ON THE ROADMAP
Relatives, other people and employers told apart from the patientSignaturesFace imagesBiometric identifiersText in images
02

Program and payer identifiers

Public payer and provider identifiers, checked against their published formats and check digits.

DETECTED
Medicare Beneficiary IdentifiersLegacy Health Insurance Claim NumbersMedicaid and CHIP member IDsHealth plan member numbersClaim and prior authorization numbersNPIDEA numbersTaxonomy codes
03

Clinical and claims code systems

Codes already written in your data.

DETECTED
ICD-10-CMICD-10-PCSICD-9-CMCPTHCPCSCDTNDCRxNormLOINCSNOMED CTCVXMS-DRGAPCCMS-HCCPlace of serviceType of billRevenue codesAdjustment and remark codesC-CDA document and section codes
ON THE ROADMAP
Suggesting ICD-10-CM, RxNorm, LOINC, SNOMED CT and HCPCS codes for facts found in free text
04

Medicaid eligibility data

Where eligibility detail sits across your data before it is shared.

DETECTED
T-MSIS eligibility groupDual eligibility codesRestricted benefit codes
05

Payment and financial data

Card, bank and credential data in billing records.

DETECTED
Card numbersExpiration datesSecurity codesPINsBank account numbersRouting numbersInternational account and bank codesTransaction IDsAmountsPasswords and other secrets

Where Safe Harbor sets a specific rule, such as keeping only the year of a date or the first three digits of a ZIP code, the engine generalizes rather than simply masking. Payment cards, routing numbers, NPIs and DEA numbers are mathematically validated, and MBIs are checked against their published format, not just pattern-matched.

Regulatory alignment

Built around the rules your data lives under

Every finding is tagged with the regulatory regimes mapped to it, so privacy, compliance and data teams work from the same map.

HIPAA de-identificationSupports both Safe Harbor and expert determination pathways.
42 CFR Part 2Designed for workflows involving substance use disorder treatment records. On the roadmap: likely Part 2 content flagged and held for your review.
Medicaid confidentiality (42 CFR Part 431, Subpart F)Designed for the safeguarding requirements that apply to Medicaid applicant and beneficiary information.
PCI DSSIdentifies cardholder data and sensitive authentication data in health and billing records.
GLBAIdentifies financial account information subject to federal financial privacy rules.
State health privacy lawsClassification designed to support state-level health and consumer privacy requirements.

Tidegate provides technology that supports privacy programs. Responsibility for regulatory determinations remains with the data holder and, where applicable, an independent expert.

One person, four records.

Step through what the engine does with one person’s records from four sources.

SOURCE RECORDS · FICTIONAL PATIENT
{{ r.src }} {{ r.badge }}
{{ c.v }}
{{ p.t }}
ENGINE OUTPUTSTEP {{ anat.n }} OF 04
{{ anat.title }}{{ anat.sum }}
{{ r.k }}{{ r.v }}
HOW
{{ h.k }}. {{ h.v }}
What every output carries

Proof that travels with the data.

01
Labels What each field is, which rules govern it and what it may be used for.
PHI · HIPAARESEARCH: YESMODEL TRAINING: NO
02
A linkage report Match quality within your data, reported by patient group in every release.
MATCH QUALITYMISSED LINKSBY PATIENT GROUP
03
A risk summary A re-identification risk summary with every release.
RISK SUMMARYEVERY RELEASE
04
Explained matches For public programs, every match is explained and logged, and no match can reduce or end coverage without a caseworker's decision.
EXPLAINEDPERSON DECIDES
Where it runs

Wherever your data already is.

OPTION 01AVAILABLEPrivate cloudIn a cloud account you control, under your keys. Deployed and operated with Tidegate Forward Deployed Engineers.
YOUR ACCOUNTSIGNED IMAGES
OPTION 02FROM V1.1On premisesOn your own hardware, in your data center.
YOUR HARDWAREYOUR KEYS
OPTION 03FROM V1.1Air-gappedWith no outside connection.
NO NETWORKSIGNED OFFLINE UPDATES

Delivered with Tidegate Forward Deployed Engineers today; packaged installs from v1.1.

Models run locallyNo calls to outside AI servicesNothing phones home

Deployed with Tidegate Forward Deployed Engineers, under a BAA whenever they access PHI.

Forward Deployed Engineers
FOR ENGINEERS AND SECURITY REVIEWERSTechnical detailsFormats, the review threshold, code examples, the release log and how the engine is tested.Read the details