To verify AI generated ABA clinical summaries against source records, break the output into material claims and trace each one to a current attributable source. Recalculate counts and denominators, check definitions, periods, settings, ordinary supports, missing data, contradictions, client and caregiver attribution, uncertainty, and corrections. A qualified clinician interprets the evidence. Keep the summary draft until every consequential claim is supported or clearly labeled as unresolved.

Define Liora's generated-summary claim ledger

Liora treats the summary as a derivative index, not a replacement chart. She maps observations, reports, calculations, model inferences, clinician interpretations, client priorities, payer actions, and open questions separately. The unit identifies approved use, client and event, source records, model and version, vendor, user and accountable author, output state, reviewer, downstream use, exception, and evidence needed for release or closure.

Build Liora's page-specific control record

Liora records summary purpose and audience, client and period, model and version, prompt and retrieval configuration, source inventory and versions, claim identifier, claim type, cited source, exact supporting span or calculation, definition, numerator and denominator, exclusions, support context, contradiction, missing evidence, client or family attribution, uncertainty, clinical reviewer, accepted text, edited text, rejected text, accessible rendering, release, correction, recipient, downstream decision, and validation. Claims with no source remain blocked or explicitly unresolved.

Put Liora's human-review boundary into practice

Liora locks the source set before review and records later additions separately. A verifier recalculates quantitative claims and compares narrative language with actual trend, variability, opportunity exposure, prompts, supports, and setting changes. The review checks whether a low count was translated into never, whether a before-and-after pattern became a causal claim, and whether omitted dissent or health context changed meaning. Client statements retain the person's communication form and are not merged with caregiver or clinician interpretations. A qualified clinician decides what the evidence means for care and explains uncertainty accessibly. If the tool summarizes an incomplete chart, the output lists missing record classes and stops short of a conclusion. Corrections update the source first, then regenerate or amend the derivative with a new version. Liora verifies every known recipient and decision that used an earlier summary.

Protect client communication and ordinary access for Liora

Liora preserves the client's direct communication, AAC, language and disability access, consent and assent when applicable, dissent, health, safety, privacy, priorities, and correction route. AI use never makes communication, food, water, bathroom access, mobility, prescribed care, rest, or emergency help conditional on tool participation or task performance.

Work through Liora's fictional example

Liora audits 28 summaries containing 210 material claims. Twenty-two summaries have full claim support. Six contain nine issues: two denominator errors, two omitted contradictions, one false causal statement, one wrong-client sentence, one stale plan, one missing dissent, and one payer-clinical conflation. The cohort teaches AI governance and denominator discipline. It does not establish treatment effect, model safety, legal compliance, coding correctness, payer acceptance, accessibility, or product performance.

Keep Liora's denominator tied to the locked population

Summary-level integrity is 22 of 28, or 78.6%. Claim-level support is 201 of 210, or 95.7%. The rates answer different questions. Corrected summaries stay in the original cohort and receive a later validated state.

Assign Liora's decisions to accountable people

Source authors own facts. Clients and families own attributed reports. Qualified clinicians interpret evidence. Reviewers verify claims. Payers own coverage actions. The model has no authority to resolve contradiction or decide care.

Address Liora's main AI documentation risk

High claim-level accuracy can conceal one severe wrong-client or safety error. Report severity and affected decisions alongside aggregate rates.

Test Liora's workflow with difficult cases

Liora tests small denominators, missing days, changed definitions, corrections, stale plans, dissent, client and caregiver attribution, payer status, causal language, wrong client, and accessible explanation.

Check Liora's release evidence

Liora confirms the exact source set and versions, client and encounter, model and configuration, approved data route, generated draft, material edits, author and reviewer decisions, accessible client communication, release destination, correction path, monitoring cohort, and known limitations. The generated-summary claim ledger retains unresolved work, owner, deadline, downstream trace, and the next revalidation trigger.

Use Liora's ABA governance sources within their scope

Liora uses the CASP public overview only for high-level organizational context. The BACB Ethics Code applies to BCBA and BCaBA certificants and applicants as defined by the Code; BACB has no separate jurisdiction over organizations or corporations. These sources support competence, documentation, confidentiality, client involvement, assessment, intervention, risk, supervision, and correction boundaries. They do not approve a tool or transfer clinical authority to software.

Keep Liora's source record and medical-review boundary visible

Liora uses current CMS Program Integrity Manual Chapter 3 as Medicare medical-review guidance. It currently says services are expected to be documented when rendered; delayed or corrected entries may occur; date and author should be identifiable; and changes or addenda clearly and permanently noted. AI output cannot supply facts that were not documented, and Medicare guidance does not become a universal payer, state, or AI rule.

Map Liora's privacy, security, and vendor roles

Liora uses the current HHS Security Rule overview, HHS cloud guidance, and HHS business-associate guidance to analyze actual covered-entity, business-associate, subcontractor, cloud, and security roles. A BAA or vendor certification does not complete purpose, permissible-use, minimum-data, configuration, risk analysis, access, incident, retention, and shared-responsibility work.

Use Liora's AI frameworks as voluntary risk tools

Liora treats the NIST AI Risk Management Framework and NIST Generative AI Profile as voluntary risk-management resources, not clinical, legal, coding, or payer authority. The profile helps teams examine generative-AI risks and actions. A framework, score, benchmark, or vendor evaluation does not prove safety, accuracy, fairness, accessibility, compliance, or fitness for this ABA use.

Protect Liora's data-purpose and communication boundaries

Liora uses HHS de-identification guidance for its two HIPAA methods and residual-risk boundary. Calling output synthetic or removing names is not itself a method. FTC staff guidance warns AI companies to honor privacy and confidentiality commitments. The OIG GCPG is voluntary and nonbinding. ASHA's AAC portal says AAC users should always have access to their tools or devices.

Choose Liora's next review trigger

Liora reopens the generated-summary claim ledger after a model, prompt, retrieval, source, template, vendor, subprocessor, setting, language, client communication method, access role, data term, payer rule, incident, complaint, correction, audit finding, or regulation changes. The review records affected people and records, immediate safeguard, owner, deadline, communication, correction, downstream propagation, and validation.

Close Liora's workflow without hiding uncertainty

Review the generated-summary claim ledger with affected clients and authorized people, qualified clinicians, health-information, privacy, security, AI-governance, accessibility, language, payer, coding, and technical leaders, and the specialists named in the manifest. Confirm source support, authorship, authority, client access, model provenance, vendor terms, errors, downstream use, correction, and independent validation. Keep this page draft and noindex until every required external review is complete.

Related resources

Sources