{"schema_version":1,"research_id":"eoa_inverse_innovation_exp05_external_evaluation_20260803","source_assessment_id":"computability_boundary_mapping__veterinary_medicine:P3:v0","cell_id":"computability_boundary_mapping__veterinary_medicine","search_queries":["site:aavld.org accreditation requirements laboratory information management system validation veterinary diagnostic laboratory PDF","site:woah.org veterinary laboratories quality standard guidelines information management software validation PDF","veterinary diagnostic laboratory errors report interpretation study laboratory information system","program equivalence undecidable arbitrary programs Rice theorem official course notes","Alive2 bounded translation validation LLVM paper PLDI 2021 equivalence checking","translation validation compiler equivalence proof certificate research paper","FDA general principles software validation automated testing migration official guidance","site:aphis.usda.gov NAHLN laboratory information management system requirements messaging veterinary diagnostic laboratories","site:fda.gov Vet-LIRN laboratory information management system validation accreditation veterinary labs","AAVLD Requirements 2024 accredited veterinary medical diagnostic laboratory information management systems software validation","Henry Gordon Rice Classes of recursively enumerable sets and their decision problems 1953 pdf","DOI 10.1090/S0002-9947-1953-0053041-6 Rice theorem"],"sources":[{"source_id":"S1","title":"An error management system in a veterinary clinical laboratory","publisher":"Journal of Veterinary Diagnostic Investigation / SAGE Publications","url":"https://journals.sagepub.com/doi/10.1177/1040638712441782","source_class":"PRIMARY_RESEARCH","publication_date":"2012-05","accessed_at":"2026-08-03","claims_supported":["Veterinary laboratory errors visibly occur across preanalytical, analytical, and postanalytical stages.","Annual recorded error rates in the studied laboratory ranged from 1.3% to 0.7% of samples during 2003–2010.","Laboratory directors reviewed error reports and corrective or preventive actions, supporting their role as operational authorizers.","An error-management system was associated with declining recorded error frequency, showing that explicit quality controls can matter."]},{"source_id":"S2","title":"A proposal to leverage high-quality veterinary diagnostic laboratory large data streams for animal health, public health, and One Health","publisher":"Journal of Veterinary Diagnostic Investigation / SAGE Publications","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC8120079/","source_class":"PRIMARY_RESEARCH","publication_date":"2021-05","accessed_at":"2026-08-03","claims_supported":["Approximately 60 accredited veterinary diagnostic laboratories used 13 veterinary LIMS products, evidencing heterogeneous deployed data pipelines.","Different nomenclatures and LIMS configurations impede aggregation and require tedious, error-prone transformations.","Veterinary laboratory data pipelines support surveillance and reporting functions with animal-health and public-health consequences."]},{"source_id":"S3","title":"AAVLD Accreditation Program","publisher":"American Association of Veterinary Laboratory Diagnosticians","url":"https://www.aavld.org/accreditation-program-2","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2017-11-15","accessed_at":"2026-08-03","claims_supported":["AAVLD accreditation evaluates laboratory quality-management systems and technical competence.","Required quality activities include document and record control, test validation, nonconforming-work control, corrective action, risk assessment, audits, and continual improvement.","Publicly funded full-service veterinary diagnostic laboratories are identifiable potential adopters, with peer inspectors and laboratory management providing governance."]},{"source_id":"S4","title":"NAHLN Activities – Resources & Guidance","publisher":"USDA Animal and Plant Health Inspection Service","url":"https://www.aphis.usda.gov/labs/nahln/activities","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"2026-01-13","accessed_at":"2026-08-03","claims_supported":["NAHLN Level 1 and Level 2 laboratories must have AAVLD or ISO 17025 accreditation, while other participating laboratories need consistent quality systems.","APHIS identifies accurate, consistent, quick, and secure electronic transmission of test results as a high priority.","NAHLN laboratories and their quality managers are identifiable adopters or partners with both quality and electronic-reporting obligations."]},{"source_id":"S5","title":"Manual of Diagnostic Tests and Vaccines for Terrestrial Animals, thirteenth edition 2024","publisher":"World Organisation for Animal Health","url":"https://www.woah.org/fileadmin/Home/eng/Health_standards/tahm/A_summry.htm","source_class":"STANDARD","publication_date":"2024-11-29","accessed_at":"2026-08-03","claims_supported":["WOAH publishes current standards for management and quality management in veterinary testing laboratories.","The manual separately addresses diagnostic-assay validation, high-throughput sequencing and computational genomics, and comparability after changes to validated methods.","Existing veterinary standards provide an adjacent governance framework but do not visibly prescribe proof-carrying equivalence decisions for arbitrary executable report pipelines."]},{"source_id":"S6","title":"General Principles of Software Validation: Guidance for Industry and FDA Staff","publisher":"U.S. Food and Drug Administration","url":"https://www.fda.gov/regulatory-information/search-fda-guidance-documents/general-principles-software-validation","source_class":"OFFICIAL_GUIDANCE","publication_date":"2002-01","accessed_at":"2026-08-03","claims_supported":["FDA guidance establishes lifecycle-oriented validation principles for regulated software and automated systems.","The guidance treats testing, requirements, intended use, documentation, change control, and risk as components of validation rather than treating a finite passing test set as a universal proof.","This is adjacent software-validation practice, although its stated scope is not veterinary diagnostic laboratory migration equivalence."]},{"source_id":"S7","title":"Alive2: Bounded Translation Validation for LLVM","publisher":"ACM SIGPLAN / Instituto Superior Técnico","url":"https://web.ist.utl.pt/nuno.lopes/pubs.php?id=alive2-pldi21","source_class":"PRIMARY_RESEARCH","publication_date":"2021-06","accessed_at":"2026-08-03","claims_supported":["A deployed tool can compare program transformations automatically using formal semantics and SMT solving.","Alive2 explicitly bounds loop unrolling and can miss bugs, demonstrating the need to retain bound-qualified verdicts.","The tool found 47 LLVM bugs and prompted semantic-specification corrections, showing both practical value and dependence on precise semantics.","Alive2 is close technical prior art for bounded equivalence checking but is specific to LLVM IR rather than veterinary LIMS pipelines."]},{"source_id":"S8","title":"CS 4820: Limits of Computability","publisher":"Cornell University","url":"https://www.cs.cornell.edu/courses/cs4820/2018fa/lectures/computability_2.pdf","source_class":"AUTHORITATIVE_SECONDARY","publication_date":"2018","accessed_at":"2026-08-03","claims_supported":["A decider must return a correct yes-or-no answer in finite time on every admitted input, while a recognizer may fail to terminate on negative instances.","Undecidability transfers by reducing a known undecidable problem into the target problem, making reduction direction material.","The notes prove the halting problem and nontrivial semantic program properties undecidable while distinguishing those class-wide results from bounded decidability and computational complexity.","The source supports the proposal's theoretical boundary, but not the correctness of its particular veterinary-pipeline reduction, which remains to be formally checked."]}],"problem_evidence":{"support":"MODERATE","rationale":"Veterinary laboratories demonstrably experience report-process errors, operate heterogeneous LIMS pipelines, and depend on accurate electronic reporting. Accreditation and network requirements make validation and controlled change consequential. No source directly measures erroneous approvals caused specifically by universal-equivalence claims during veterinary pipeline migrations, so the exact proposed failure mode remains an inferred intersection of documented laboratory and software risks.","source_ids":["S1","S2","S3","S4","S5","S6"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"Public veterinary diagnostic laboratories, laboratory directors, quality managers, AAVLD, and USDA NAHLN are identifiable adopters, authorizers, or institutional partners. They expressly require quality systems, validation, and accurate electronic reporting. None of the eight sources expressly requests formal equivalence certificates or an UNRESOLVED-aware migration router, leaving product-specific stakeholder pull unverified.","source_ids":["S1","S3","S4"]},"prior_art":{"proximity":"ADJACENT_PRIOR_ART","closest_analogues":[{"name":"Alive2 bounded translation validation","similarity":"Automatically checks semantic preservation between program versions using formal semantics and SMT solving, with explicit resource bounds.","remaining_difference":"It targets LLVM transformations, can miss bugs because of bounds, and does not supply veterinary report contracts, laboratory authority routing, or production fallback governance.","source_ids":["S7"]},{"name":"FDA lifecycle software validation","similarity":"Requires intended-use requirements, documented validation, testing, change control, and risk-sensitive evidence for software.","remaining_difference":"It does not provide a computability classification, independently checkable program-equivalence certificate, or explicit evidence lattice for veterinary pipeline migrations.","source_ids":["S6"]},{"name":"AAVLD/WOAH veterinary laboratory quality management","similarity":"Provides established accreditation, validation, nonconformance, audit, corrective-action, and change-comparability practices for veterinary laboratories.","remaining_difference":"The reviewed materials do not visibly define arbitrary-program equivalence, checked reductions, proof certificates, bounded exhaustive verdicts, or UNRESOLVED-preserving release routing.","source_ids":["S3","S5"]},{"name":"Differential testing and laboratory error management","similarity":"Compares observed behavior, records concrete discrepancies, and routes detected nonconformities to laboratory management.","remaining_difference":"Observed cases can refute equivalence but cannot prove termination and identical reportable output for every admitted future record.","source_ids":["S1","S6"]}],"distinctive_claim_remaining":"For a veterinary laboratory migration, a version-linked release gate can constrain its verdict to the strongest independently checked artifact available—contract-scoped termination-and-equivalence certificate, replayable counterexample, exhaustive finite-envelope result, or UNRESOLVED—and prevent timeouts, failed proof searches, or finite regression passes from authorizing unrestricted equivalence. This is contrastive and falsifiable at the workflow level; it is not a claim of world novelty, clinical correctness, or universal proof-search success.","confidence":"MODERATE"},"implementation_evidence":{"support":"MODERATE","rationale":"Computability theory supports rejecting a total exact decider for a matched unrestricted semantic class, and Alive2 demonstrates bounded automated equivalence analysis in a mature programming ecosystem. FDA, AAVLD, APHIS, and WOAH practices support requirements, validation, audit, change control, and accountable quality management. Major unresolved work includes formalizing each pipeline language and external environment, proving that the proposed reduction matches that model, capturing clinically material report semantics, integrating proprietary plugins, obtaining an independent checker and reviewer, and confirming that a laboratory can operate the four-state router without unsafe reinterpretation.","source_ids":["S3","S4","S5","S6","S7","S8"]},"scores":{"meaningful_impact":{"score":4,"rationale":"Incorrect or delayed veterinary reports and inconsistent surveillance data can affect animal care and public-health response, while explicit evidence boundaries could prevent both unsafe approval and needless migration blockage. The magnitude attributable specifically to migration-equivalence errors is not measured.","source_ids":["S1","S2","S4"]},"stakeholder_pull":{"score":3,"rationale":"Laboratories and network authorities visibly demand validation, quality systems, and accurate electronic results, but no direct demand for this formal gate was found.","source_ids":["S3","S4"]},"incremental_advantage":{"score":3,"rationale":"The proposal improves honesty and auditability beyond regression testing by separating proof, counterexample, bounded equivalence, and unresolved outcomes. Comparative accuracy, delay, and cost advantages have not been tested.","source_ids":["S6","S7","S8"]},"distinctiveness_plausibility":{"score":3,"rationale":"The veterinary-specific evidence router and authority record differ from reviewed formal-method and laboratory-quality analogues, although their constituent practices are established and the search is not a novelty review.","source_ids":["S3","S5","S7"]},"technical_implementability":{"score":3,"rationale":"Toy and restricted implementations are credible, but arbitrary plugins, external calls, numeric behavior, nondeterminism, and semantic fidelity can make real deployments difficult or perpetually unresolved.","source_ids":["S6","S7","S8"]},"adoption_authority_feasibility":{"score":3,"rationale":"Laboratory directors and accredited-laboratory quality structures provide plausible authority paths, but certificate acceptance, procurement, liability, and production release policy were not directly verified.","source_ids":["S1","S3","S4"]},"evidence_readiness":{"score":2,"rationale":"Theoretical and adjacent-practice evidence is substantial, but there is no veterinary migration dataset, prototype result, checked target reduction, laboratory interview, workflow observation, or prospective comparator study.","source_ids":["S1","S7","S8"]},"safety_net_benefit":{"score":4,"rationale":"Preserving legacy operation and explicit UNRESOLVED status could reduce false approvals, provided unresolved verdicts cannot be silently coerced and the legacy path is itself safe.","source_ids":["S3","S6","S8"]},"scalability":{"score":2,"rationale":"The governance pattern is reusable, but formal semantics, report projections, environment models, and certificates may require substantial migration-specific work; bounded tools explicitly trade coverage for feasibility.","source_ids":["S2","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"10K_TO_50K","scope":"Two synthetic pipeline pairs, a schema of at most four fields and 256 records, executable comparator, basic proof artifacts, router tests, and a second technical review.","confidence":"MODERATE","assumptions":["One software/formal-methods engineer for approximately two to four weeks.","A veterinary pathologist or laboratory-quality reviewer contributes limited contract-review time.","Open-source solver and proof-checking tools are sufficient.","No production data, clinical reports, or regulated release changes are used."],"source_ids":["S7","S8"]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"Discovery and formalization for one laboratory: report projection, input and environment contracts, code identity, threat and safety review, governance design, and adapters for one representative migration.","confidence":"LOW","assumptions":["Source code and plugin behavior are available.","One formal-methods engineer, one laboratory software engineer, and fractional pathology/quality participation work for two to six months.","Existing CI, audit logging, and identity infrastructure can be reused.","This phase does not authorize production release."],"source_ids":["S3","S6","S7"]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Production integration for one laboratory, including hardened execution isolation, pipeline adapters, artifact retention, independent review, release-system controls, validation documentation, training, and monitored dual running.","confidence":"LOW","assumptions":["One to three major pipeline families are in scope.","The laboratory retains the legacy release path during validation.","No wholesale LIMS replacement or new clinical assay validation is included.","Security, legal, procurement, and accreditation review are required."],"source_ids":["S3","S4","S6"]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Artifact storage and compute, solver and dependency maintenance, contract updates, certificate review, regression-envelope maintenance, audits, incident response, and rechecking after schema, code, plugin, or reference-data changes.","confidence":"LOW","assumptions":["A single laboratory processes a modest number of major migrations annually.","Most migrations reuse existing semantics and adapters.","Independent specialist review is fractional rather than full-time.","High unresolved rates or proprietary verification tooling could raise costs beyond this band."],"source_ids":["S3","S6","S7"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Veterinary laboratory errors, heterogeneous LIMS data transformations, and high-priority accurate electronic reporting are externally documented, although the frequency of equivalence-related migration failures is not.","source_ids":["S1","S2","S4"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"AAVLD-accredited and NAHLN laboratories, laboratory directors, and quality managers are identifiable adopters and authorizers operating under documented quality and reporting obligations.","source_ids":["S1","S3","S4"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The router's classification accuracy, preservation of bounds, refusal to approve UNRESOLVED cases, artifact traceability, reviewer agreement, time, and cost can be compared directly with regression-only review.","source_ids":["S6","S7","S8"]},"bounded_next_evidence_step":{"status":"YES","reason":"A synthetic, offline test with at most 256 inputs, two known pipeline pairs, predefined verdicts, a regression comparator, and explicit failure conditions is bounded and avoids clinical release.","source_ids":["S7","S8"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The proposed first step is offline and synthetic, does not issue reports, and can be halted if the router overstates evidence. Production authority, liability, and clinical-semantic questions remain later-stage gaps rather than stops for this experiment.","source_ids":["S1","S3","S6"]},"credible_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The scopes and staffing assumptions are bounded, but no direct vendor pricing, laboratory labor estimate, integration quotation, or observed implementation effort was found. Production ranges are therefore resource-equivalent planning estimates, not externally validated budgets.","source_ids":[]}},"next_evidence_step":"Run a preregistered offline prototype on a synthetic four-field specimen schema with no more than 256 enumerated records. Construct Pair A to be equivalent and terminating under a written contract and Pair B to differ on exactly one valid record. Compare (1) the proposed artifact gate—proof search, independent certificate checking, exhaustive finite-envelope comparison, counterexample replay, and four-state routing—with (2) a time-matched conventional regression suite sampled from the same envelope. Primary measures are correct final label, false-equivalence approvals, preservation of the finite bound in every output, reproducibility of the planted counterexample, independent-review agreement, analyst hours, compute, and elapsed time. Independently review a written halting-to-pipeline-equivalence reduction for source-to-target direction, total computability of the construction, answer preservation, and exact model match. Falsify the intervention if either known pair is misclassified, UNRESOLVED authorizes release, a timeout becomes DIFFERENT or EQUIVALENT, the finite-envelope label loses its bound, the proof checks a materially different report relation, the planted mismatch cannot be replayed, or the synthetic exercise exceeds 50K USD resource-equivalent cost without explaining a production-relevant capability. Success would establish artifact and workflow integrity only, not production feasibility, clinical benefit, unrestricted equivalence, patentability, or world novelty.","blocking_evidence":["No direct prevalence estimate for veterinary pipeline-migration equivalence failures.","No veterinary laboratory has expressed demand for proof-carrying equivalence or agreed to adopt the proposed router.","The proposed unrestricted reduction has not been formalized or independently checked against an actual pipeline model.","No operational report-equivalence contract has been validated by a veterinary clinical pathologist.","No evidence shows that proprietary plugins, external reference data, nondeterminism, floating-point behavior, and network dependencies can be modeled faithfully.","No prototype has measured false approvals, unresolved rates, reviewer burden, latency, or cost against regression testing.","Production release authority, liability allocation, cybersecurity review, procurement constraints, and accreditation treatment remain unverified.","Cost bands lack direct quotations or observed veterinary-laboratory implementation data."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"The search assessed only eight direct sources for problem visibility, stakeholder authority, adjacent practice, and feasibility. It did not measure world novelty, patentability, freedom to operate, market size, or realized impact. No claim is made that the combined protocol is globally new or absent from unpublished, proprietary, patent, or laboratory-internal systems.","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":false,"material_progress_observed":true,"progress_targets":["Complete the preregistered synthetic comparator experiment and publish all verdicts, artifacts, bounds, failures, labor, compute, and elapsed time.","Obtain independent verification of the unrestricted reduction and document every model assumption and unmatched veterinary-pipeline feature.","Secure a veterinary laboratory director and clinical pathologist as research partners and elicit a signed report-equivalence and release-authority contract.","Measure current migration practice, failure modes, approval criteria, and willingness to preserve UNRESOLVED through interviews and workflow observation.","Test one non-production representative pipeline pair with proprietary dependencies under a fixed replayable environment.","Acquire bottom-up staffing estimates, integration quotations, and recurring-maintenance estimates sufficient to validate or revise all four cost bands."],"reason":"Bounded web research supports the general problem, credible institutional actors, theoretical boundary, and adjacent validation techniques, but cannot establish veterinary-specific demand, semantic fidelity, router performance, operational unresolved rates, or real implementation cost. Those decisive gaps require interviews, artifact construction, proprietary pipeline access, and live offline testing; under the controller rule this requires an empirical-research stop."},"proposal_index":3}