How do I scope the DSI Source Attributes audit?

Use the brass 'Prepare evidence for DSI Source Attributes review' button. First evidence record within 21 days of access; re-runs in a single business day.

DSI Source Attributes.

Prepare evidence for DSI Source Attributes review

Buyer translation: this reference page maps the mark to a tool, system, control, deployment, or evidence question. the 31 predictive-DSI source attributes as an evidence checklist · DSI Source Attributes can require evidence around clinical workflows, ePHI boundaries, change control, or patient-impacting decisions. Planisphere maps the workflow, runs the harness, and packages the measurement record for internal or external compliance review; it does not certify compliance, perform the HIPAA risk analysis, or replace the safeguard system of record.

Reviewed 2026-07-02

01 ·

What this is.

Sixty-nine marks · cite-anchors for every regulatory regime Planisphere maps to · postmark idiom · single distinctive brass glyph per control · universal classifier · pSEO surface

Use. the 31 predictive-DSI source attributes as an evidence checklist.

Source. tag rows · brass checklist rail.

Register. 09 · Compliance controls — one of the ten registers of the PLANiSPHERE corpus library.

02 ·

Promise. compliance

The contract this mark binds — derived, not asserted.

DSI Source Attributes can require evidence around clinical workflows, ePHI boundaries, change control, or patient-impacting decisions. Planisphere maps the workflow, runs the harness, and packages the measurement record for internal or external compliance review; it does not certify compliance, perform the HIPAA risk analysis, or replace the safeguard system of record.

Answers: What evidence can Planisphere prepare for DSI Source Attributes review?

03 ·

See it work.

Evidence record you can check — not code you have to trust.

Planisphere measures your tool deployment against DSI Source Attributes and seals the result into a signed, Merkle-rooted evidence record. The grade recomputes on your own hardware; the model state never crosses the boundary.

Prepare evidence for clinical AI review →See how a record is checked

04 ·

What it requires.

The obligation this regime imposes — and the honest ceiling. Healthcare regulation

The DSI Source Attributes are the operational checklist behind HTI-1: the 31 attributes a Predictive DSI must declare (13 for evidence-based DSI). Several are measurable properties of model behaviour rather than paperwork — output description, intended decision-making role, validation cohort, fairness/FAVES posture, and ongoing performance. ONC's intent is that these attributes be true and current — an industry-standard floor — which is a measurement claim that decays with drift.

What the evidence record proves.

Distinctness across declared cohorts backs the fairness/FAVES rows
Drift delta against the declared baseline backs the 'ongoing performance' row
Coherence score backs the 'intended decision-making role' row
A per-attribute SHA binds each measurable attribute to one audit axis
The record is the evidence layer under the developer's §170.402(b) maintenance duty

What this does not prove. Not every source attribute is measurable — provenance, funding source, and developer identity are declarations, not measurements. Planisphere attests the measurable subset and leaves the declarative subset to the developer; no PHI, no certification, not legal advice.

Cite-anchor: ONC HTI-1 — DSI source attributes (31 predictive / 13 evidence-based) · 45 CFR 170.315(b)(11)(iv)–(v); maintenance under 45 CFR 170.402(b)

05 ·

Questions this answers.

2 high-intent questions buyers search — each on its own page, each pinned to DSI Source Attributes.

What are the 31 source attributes required for predictive DSI under HTI-1? Read the answer → Do I have to keep HTI-1 source attributes updated, and how often? Read the answer →

06 ·

The board that carries this.

Related marks, and the surface this one funnels to.

DSI Source Attributes asks for an evidence record: The DSI Source Attributes are the operational checklist behind HTI-1: the 31 attributes a Predictive DSI must declare (13 for evidence-based DSI). The clinical AI workflow produces that record — see how it runs at /medicine.

CMS reg 09 · Medicare/Medicaid coverage · algorithm cannot be the sole basis to deny care FDA PCCP reg 09 · Predetermined Change Control Plan · drift-bounded model updates FDA PCCP / SaMD reg 09 · Predetermined Change Control · medical AI

07 ·