Balsam Semantica Computable clinical meaning

The instrument

Clinical Intelligence Readiness Instrument (CIRI™)

An independent, physician-led assessment of the institutional conditions required to govern clinical AI.
CIRI™ asks three questions of an institution: whether it can preserve clinical meaning in machine-readable data, whether its clinicians retain authorship over decisions influenced by AI, and whether it can account for the systems operating in its clinical environment.
It does not rely on what an institution says about itself. It weighs documented evidence, and compares independent accounts from the people responsible for governance, clinical practice, and technical infrastructure.
18 metrics. 3 dimensions. 3 independent respondent groups.

The result is a structured institutional position, not a certification or a prediction of AI safety.

18metrics across three conditions
3respondent groups, answering independently
0–5scale, evidenced above the second level
10working days from a complete submission

Who this is for

For Institutions Accountable for Clinical AI

CIRI™ is designed for healthcare institutions preparing to deploy clinical AI, and for those that want to know whether their existing governance holds up in practice.Its intended readers include:
• Chief Medical Information Officers
• Chief Information Officers
• Chief Risk and Governance Officers
• Clinical AI governance committees

What it measures

Three Dimensions of Institutional Readiness

Weight of the index

Semantic Sovereignty

Whether the institution governs the meaning of its own clinical data: how it is documented, how it is coded, what survives when it moves, and whether the reasoning behind a decision is still in the record at all.

Weight of the index

Clinical Authorship

Whether clinicians hold real authority over decisions systems influence: in the room where deployment is decided, at the moment output appears, and after the fact when someone is answerable.

Weight of the index

Infrastructural Readiness

Whether the institution can enumerate the systems operating within it, monitor them, reconstruct what they did, and withdraw one without withdrawing the care it supports.

Semantic Sovereignty40%

Clinical Authorship30%

Infrastructural Readiness30%

Each dimension carries six metrics, eighteen in total.

EVIDENCE OVER SELF-REPORTING

What Exists, What Can Be Shown, and What Is Unaccounted For

CIRI™ draws on three respondent groups:
• Governance: the officers responsible for policy, oversight, and contractual obligations
• Clinical Practice: clinicians working at the point of care

• Technical Infrastructure: the professionals responsible for systems, data models, and terminology mappings.

Each metric is scored from 0 to 5 against predefined conditions. Any score above level 2 must be supported by institutional evidence: specifications, query results, audit records, or contractual provisions.
Responses are collected independently and are not reconciled beforehand.

Governance account
4
Clinical account
2
Divergence Index
2

Illustrative. The lower answer sets the score; the distance between the two accounts is reported alongside it.

What you receive

An Evidence-Based Institutional Position

  • Each assessment produces a written report containing:
    • The overall CIRI™ score out of 100 • Scores for each of the three dimensions • The Divergence Index • Individual scores for all eighteen metrics • The evidence supporting or limiting each score • The conditions required to reach the next level
    The report states where the institution stands. It does not prescribe a remediation programme or replace specialist implementation work.

THE DIVERGENCE INDEX

The Distance Between What an Institution Has Decided and What It Does

Ten of the eighteen metrics are answered by two respondent groups independently. Where their accounts differ, the lower response sets the score, and the gap is reported separately as the Divergence Index.
A strong institutional score does not guarantee consistent practice. The Divergence Index shows where the formal arrangement and the working reality part ways.

How it works

Three Independent Accounts, One Consolidated Report

  1. The institution names a coordinator. Accountable for distributing the forms, collecting responses, and submitting artefacts. 
  2. The forms are completed independently and returned with their artefacts. Most institutions take two to three weeks. There is no deadline, the assessment begins when the submission is complete.
  3. The report is delivered within ten working days of a complete submission.

Assessment windows. Assessments are conducted in defined windows during the year, published in advance. A request received between windows is held and confirmed when the next opens. This is what makes a fixed delivery period possible: an assessor accepting work continuously cannot promise a date to anyone.

The assessment is cross-border by design. It is delivered in writing, on the institution's own time.

The institution names a coordinator, accountable for distributing the forms, collecting responses, and submitting artefacts.
The forms are completed independently and returned with their artefacts.Most institutions take two to three weeks. There is no deadline; the assessment begins when the submission is complete.
The report is delivered within ten working days of a complete submission.

The institution names a coordinator, accountable for distributing the forms, collecting responses, and submitting artefacts.The forms are completed independently and returned with their artefacts.Most institutions take two to three weeks. There is no deadline; the assessment begins when the submission is complete.The report is delivered within ten working days of a complete submission.

What the institution must accept before commissioning

Before the assessment begins

These conditions :

The three groups answer independently

Responses are not circulated or reconciled before submission. The Clinical Practice group requires at least three clinicians, and the executive sponsor may not be one of them.

Clinicians are reached directly

The coordinator supplies names and email addresses; the forms go out and come back without passing through the institution. No answer is attributed by name. A clinician who expects the executive to read their answers is not answering independently, and the comparison the assessment exists to make depends on that independence..

Artefacts are submitted with the responses

Any claim above the second level with no artefact attached is recorded at the level it can evidence, without correspondence. A certificate of conformance to a standard is not accepted in place of a requested artefact.

The report is yours to use

A perpetual licence to use it internally, with your board, with your regulators and with your suppliers. Copyright in the report and in the instrument remains with Balsam Semantica, and the report is not published in full or used to derive an instrument of its own.

Divergence is an expected result, not an accusation

The signatory is ordinarily the governance track, and it is the governance account the index most often qualifies. That is the assessment working, not failing.

An incomplete submission produces a completed assessment

Where artefacts are withheld, or a track is silent, the report is delivered with the score recorded as not assessable and the withholding named. This is a finding and a full deliverable, and it does not entitle the institution to a refund.

A submission can be completed for twelve months, at no further cost

If a track responds late, or an artefact is located after delivery, the missing material is submitted and the assessment is reissued in full - no second purchase, and no repeat of the questionnaire. Nothing that was already answered is answered again. An institution therefore has a year to complete what it began, and no reason to withhold anything permanently.

Why the assessor is not in the room

A condition of the method, not a preference

The assessor is not present while the institution reads the finding. That is a condition of the method, not a preference of the assessor's.

An assessment that is presented is an assessment that is adjusted. The room does its work - a raised eyebrow at a score, a question asked twice, and the reading softens by a degree. This is not a failure of will; it is what presence does, and it is why an auditor's independence is protected by distance rather than by resolve.

The report is therefore written to be read without its author. Every paragraph in it exists before the institution does: the scale, the wording at each level, the condition above it. Nothing is composed in response to who is reading, and nothing can be revised by asking.

What can be asked is whether the instrument was applied correctly. That question is answered in writing, from the record, and the answer is the same whoever asks it.

A sample is available

A single metric, taken from a worked assessment and stripped of any identifying detail, showing the score, the reason it was limited, and the condition required to reach the next level. It is the whole shape of the report in one page. See the sample metric.

What this is not

Conditions prior to performance, not a substitute for it

This is not a certification, an accreditation, a safety assurance, or a regulatory approval, and it must not be presented as any of those. It does not assess the clinical performance of any system.

It assesses the conditions under which a system's output can be governed - conditions that are prior to performance, not a substitute for it.

No commercial relationship with any supplier

Balsam Semantica holds no partnership, reseller agreement, referral arrangement or financial interest with any vendor of clinical systems, terminology services or AI products. No supplier pays to be assessed favourably, and none could. The instrument does not evaluate, rank or recommend products, and no finding names a supplier as preferred.

The report states a position. It does not recommend a course of action: how an institution should close a gap, in what order and at what cost depend on circumstances the instrument does not collect, and an assessment that guessed at them would be advising rather than measuring.

Fee

Fee is confirmed with the opening of the next assessment window.

One fee, one scope.

Payment is processed by our payment provider, which acts as the seller of record and issues the invoice.

Request the fee and terms →

[email protected]

Versions

Issued once a year

CIRI™ is issued in versions, once a year.

The conditions it assesses are moving. Documentation composed by systems, capability arriving inside software bought for other purposes, meaning transformed at speeds no one reviews - none of these existed in their present form when the first version was written, and none will hold that form for long. An instrument that measured the field once would describe it accurately for about a year.

Between versions the instrument does not change: two institutions assessed in the same year are assessed identically, and every score carries the version it was measured on. An institution assessed on one version can be assessed again on the next, and the distance between them is a finding in itself.

Who conducts it

A clinician's judgement, applied to semantic evidence

CIRI™ was developed and is conducted by Dr Balsam Khalil, MD, a physician whose work focuses on computable clinical meaning: what is preserved, altered or lost when the meaning of a clinical encounter becomes structured information that systems interpret and act upon.The assessment is grounded in clinical judgement. Reading a clinical record requires more than identifying terminology or checking structural completeness. It requires understanding what was established during the encounter, what remains uncertain, and what may have been lost in its representation. This foundation comes from medical training and clinical reasoning.
Her work across clinical documentation and Revenue Cycle Management has brought into focus how clinical meaning travels from the consultation through documentation, coding and data exchange, and how it can change at each stage. 
The instrument does not produce the assessment; it structures it. What is evaluated, and why, remains a matter of clinical judgement. A specification is examined for whether its requirements are governed or merely documented. A contractual provision is considered for its implications when put into practice. A terminology mapping is evaluated for what it preserves, alters or loses in meaning, not simply for whether it conforms to a technical specification.
The instrument provides the structure.
The judgement remains human, independent and accountable.

CIRI™ - the Clinical Intelligence Readiness Instrument 

Questions and answers about the assessment

Everything institutions usually want to know before commissioning an assessment. If something is not here, write and ask.

Who carries out the assessment?

Dr Balsam Khalil, a physician, reviews all submitted documents and makes every assessment decision. No part of the assessment is delegated or automatically generated.Clinical training is essential to understanding what the documents actually show. For example, whether a rationale field contains clinical reasoning or simply repeats a diagnosis. Whether a terminology mapping preserves the original clinical meaning or changes it. And whether missing information means a normal finding, a question that was never asked, or a field that had no place to record the answer.

How do we communicate during the assessment?

All communication takes place in writing through one agreed channel. This keeps a clear record of decisions and means that no one needs to be available at a specific time.It also protects the independence of the assessment. Findings should not change through informal discussions or pressure during the review. The scoring criteria and their interpretation are defined in the instrument before the assessment begins.If you have a question about a finding or believe a criterion has been applied incorrectly, you can request a written explanation based on the assessment evidence.

How does payment work?

The fee is fixed, communicated in advance, and the same for every institution. Full payment is required before the assessment begins.This ensures that the assessment is not influenced by whether its findings are welcome to the institution.Before payment, you will know exactly what is included, how much it costs, and when the report will be delivered. The scope, fee and delivery period are all agreed in advance.

What if we cannot provide all the requested documents?

The assessment can still proceed, and you will receive a full report.When evidence is missing, the report identifies what was not provided and explains which metrics are affected. Scores are based on available evidence, not assumptions.You then have twelve months to submit the missing documents at no additional cost. Once they are received, the assessment is updated and reissued in full. You will not need to repeat information already submitted.Missing evidence is not unusual. It can also reveal gaps in documentation, oversight or institutional processes.

Why do clinicians answer separately?

Because comparing institutional policy with actual clinical practice is an important part of the assessment.Ten of the eighteen metrics are answered independently by two groups: those responsible for setting institutional policies and those who work under them.Differences between their answers can reveal gaps between what an institution has established and what happens in practice. These differences may be lost if answers are shared or discussed before submission.Clinicians therefore receive and return their forms directly. Their individual answers are confidential and are not attributed to them by name.This independent process is a required part of CIRI™.

What is the Divergence Index?

The Divergence Index measures the gap between institutional policy and clinical practice.Ten metrics are answered independently by two groups. When their answers differ, the lower score determines the metric's final score. The difference between the two scores is also reported separately as the Divergence Index.These two figures provide different information.A high score with high divergence may indicate that institutional policies are well developed but are not consistently reflected in practice.A moderate score with low divergence may indicate that institutional policies and clinical practice are more closely aligned, even if further improvements are needed.

Does the report include recommendations?

The report shows your current level for each metric and explains what is required to reach the next level. It does not provide an implementation plan.This is intentional. The right actions depend on your budget, systems, staff and existing processes. CIRI™ does not collect enough information about these factors to prescribe specific solutions responsibly.Instead, you receive a documented assessment that your internal teams or external advisors can use to plan improvements.

Is this a certification?

No. CIRI™ is an independent assessment, not a certification.Each report is signed, dated and linked to a specific version of the instrument.CIRI™ does not evaluate the clinical performance of an AI system. It evaluates the institutional conditions needed to govern clinical meaning when health information is used by AI and other digital systems.

Do we need to be using SNOMED CT or FHIR?

No. The instrument is standards-agnostic and does not test conformance to any of them.

It asks what your mappings do to clinical meaning, whichever standard sits underneath. An institution running a local terminology can score well if it governs what that terminology does to meaning. An institution fully conformant to an international standard can score less well, because naming a standard in a requirement is the lowest level this instrument recognises - what matters is how it is used, bound, versioned and maintained.

Is CIRI™ a governance framework?

No. A governance framework describes what an institution should establish. CIRI™ assesses what is actually in place and whether the available evidence supports it.Implementing improvements identified in the report is separate work. Institutions can undertake this internally or engage another provider.Balsam Semantica does not provide implementation services as part of CIRI™. This separation helps preserve the independence of the assessment

How is this different from a readiness self-assessment?

A self-assessment usually asks an institution to rate its own readiness. CIRI™ examines documented evidence and uses it to determine the score.Questions focus on what exists, what has been done and what can be demonstrated. Higher assessment levels require supporting evidence, such as specifications, query results, audit records or contract clauses.CIRI™ also collects independent responses from three groups: institutional leadership, clinical staff and technical or operational personnel.The differences between their responses are measured and reported. A self-assessment completed by one person cannot provide this comparison.

How does this compare with a consulting firm's review?

They serve different purposes, and a consulting firm may be the better choice for some projects.Consulting firms typically review several areas, including infrastructure, cybersecurity, operational processes and organisational maturity. Their work often involves a team, workshops and interviews.CIRI™ has a narrower focus: assessing whether clinical meaning is accurately represented, preserved and governed as information moves from clinical documentation into digital systems.The assessment is also designed to require less staff time. It involves three independent forms, supporting documents and a final report, without workshops or on-site teams.If you need organisation-wide transformation or implementation support, a consulting firm may be more suitable.If you need an independent, evidence-based assessment of how well your institution governs clinical meaning before deploying AI, CIRI™ addresses that question.

Do you have relationships with any of the vendors we use?

No. Balsam Semantica has no partnerships, reseller agreements, referral arrangements or financial interests with suppliers of clinical systems, terminology services or AI products.The assessment examines how your institution manages its suppliers. This includes contractual responsibilities, ownership and maintenance of terminology mappings, and notification of changes that may affect clinical meaning.It does not evaluate whether you have selected the right supplier, and it does not recommend replacing one.

Does it look at vendor lock-in?

Yes. CIRI™ examines whether an institution can access and retain control of its clinical information and terminology mappings independently of its suppliers.This includes questions such as:Can you access and use your terminology mappings outside your supplier's system?Does your institution control the information model used to structure its clinical data?Are you notified in advance when terminology updates may affect clinical meaning?Have your data and mappings actually been exported and tested, or is portability only promised in theory?An institution may be contractually free to leave a supplier but still depend on that supplier to understand or use its own clinical data.CIRI™ helps identify this risk before a contract is renewed or a system is changed.

What happens to our documents?

Your documents and assessment report remain confidential. Nothing is shared with third parties without your written permission.CIRI™ does not require identifiable patient information. Please do not submit patient data.The assessment focuses on institutional documentation, such as technical specifications, terminology mappings, policies, contracts and audit records.Anonymised, aggregated scores may be retained to help improve and calibrate future versions of the instrument. No institution-identifying information is retained for this purpose.

When can an assessment take place?

Assessments are conducted during scheduled periods throughout the year. These periods are published in advance.Requests received between assessment periods are recorded and confirmed when the next period opens.Working within scheduled periods allows delivery dates to be agreed in advance and applied consistently to all participating institutions.

Can we be reassessed later?

Yes, and it is worth doing.

CIRI™ is issued in an annual version and remains unchanged throughout that assessment year. This ensures that institutions assessed during the same year are evaluated using the same criteria.A reassessment allows you to compare changes in individual metric scores and in the Divergence Index.This comparison can show whether institutional improvements have led to better alignment between policies and actual clinical practice.

Clinical meaning is not generated by AI. It is captured, governed, and made computable.

Clinical Intelligence Readiness Instrument  (CIRI™) - proprietary instrument. Licensed, never assigned.

Something not answered here? Write, and you will get a straight answer: [email protected]