{"schema_version":1,"assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","source_experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"invariant_mode_decomposition_design__neuroscience","archetype_slug":"invariant_mode_decomposition_design","domain_slug":"neuroscience","title":"Retrospective testing of distributed preictal modes for seizure warning and stimulation targeting","opportunity_summary":"Test whether a locally estimated multielectrode transition mode provides held-out seizure warning and stimulation-response information beyond channel-wise monitoring and a nonlinear multichannel rival. The proposed first step is retrospective only; prevalence, novelty, transferability, and causal benefit remain unestablished.","adopter_authorizer":"For the first study: a single-site epilepsy-monitoring program, its data custodian and investigators, and the responsible ethics/IRB authority. Any prospective stimulation would additionally require patient consent and authorization from the treating epilepsy team.","scores":{"meaningful_impact":{"score":4,"rationale":"If the hypothesized missed distributed instability exists, earlier warning and better-targeted stimulation could reduce seizure progression, ineffective stimulation, and neural-tissue exposure. The affected outcomes are important, but the frequency and magnitude of the proposed failure are unsupported in the sealed candidate."},"stakeholder_pull":{"score":3,"rationale":"The proposal addresses recognizable concerns of patients, epilepsy clinicians, and investigators and names their roles, but it supplies no evidence that these stakeholders currently demand a modal approach or regard channel-wise monitoring as a material unmet problem."},"incremental_advantage":{"score":3,"rationale":"The candidate specifies testable additions over channel-wise thresholds and a nonlinear predictor: stable modal gain, mode identity, and a dynamical stimulation map. Whether these improve warning or response prediction is wholly untested and could be negated by either comparator."},"distinctiveness_plausibility":{"score":3,"rationale":"Linking a distributed weakly damped mode to both warning and contact-pattern selection is proposal-specific and experimentally separable from its stated rivals. Prior art is explicitly unsearched, so distinctiveness in the external field cannot be credited."},"technical_implementability":{"score":3,"rationale":"A retrospective replay using existing multielectrode recordings and stimulation logs is technically bounded, with patient-level splits and measurable residual, gap, and drift gates. Nonlinearity, partial observation, non-normality, mode swapping, artifacts, and retrospective confounding create substantial feasibility risk."},"adoption_authority_feasibility":{"score":3,"rationale":"The candidate clearly identifies data-custodian, investigator, IRB, consent, and treating-team authority and keeps the first test noninterventional. Moving from retrospective evidence to stimulation decisions would face materially greater clinical, ethical, and likely compliance requirements that are not resolved here."},"evidence_readiness":{"score":3,"rationale":"The authorized first study can use existing intracranial recordings and clinically delivered stimulation logs, and the candidate prespecifies comparators and falsifiers. Actual data quantity, seizure counts, stimulation variation, subgroup coverage, data quality, and access are not established."},"safety_net_benefit":{"score":4,"rationale":"The retrospective first step changes no care, while residual, false-alarm, spectral-gap, drift, and subgroup gates provide explicit reasons to halt claims and revert to the ordinary baseline. These safeguards do not resolve risks of false reassurance or unsafe prospective control."},"scalability":{"score":2,"rationale":"The approach depends on invasive monitoring, patient-specific electrode coverage, locally valid transition estimates, and repeated validation across patients and recording regimes. The candidate itself identifies poor transfer and missing causal tissue as risks, and provides no evidence of cross-site or cross-patient scalability."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"One single-site retrospective replay using existing intracranial recordings and stimulation logs, including data governance, IRB preparation, preprocessing, model implementation, comparator evaluation, patient-level validation, subgroup reporting, and clinical review.","confidence":"LOW","assumptions":["Existing recordings and stimulation logs are already held by the partner site and do not require new acquisition.","The dataset is sufficiently organized for analysis without extensive manual reconstruction.","The study uses existing computing infrastructure and a small multidisciplinary team.","No prospective stimulation, treatment change, or device integration is included."]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Preparation of a single-site prospective shadow-mode system, including robust software, streaming integration, monitoring dashboards, protocol and compliance work, cybersecurity review, usability testing, and prospective observational validation without directing stimulation.","confidence":"LOW","assumptions":["The system initially runs silently or as research-only decision support.","Existing electrophysiology acquisition infrastructure exposes usable data interfaces.","The estimate excludes development of new implantable hardware.","Substantial site-specific engineering and clinical coordination are required."]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"A limited clinical launch at one or a few epilepsy centers involving validated real-time operation and, only after separate authorization, integration into clinician-supervised stimulation decisions, with quality systems, training, safety monitoring, and formal evaluation.","confidence":"LOW","assumptions":["Launch reuses clinically available electrodes and stimulation equipment.","Prospective intervention requires additional consent, clinical governance, and compliance work.","The scope is limited-center launch rather than broad commercialization.","Unexpected device-development or pivotal-trial requirements could move the cost above this band."]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Recurring support for a limited multi-center operation, including data pipelines, model monitoring, site support, clinical oversight, software maintenance, audit and compliance activities, subgroup surveillance, and periodic revalidation.","confidence":"LOW","assumptions":["Only a small number of centers are supported.","Existing clinical hardware remains available.","Patient-specific validation and monitoring remain necessary.","The band excludes major new trials, device replacement, and broad commercial field operations."]}},"research_burden":"HIGH","earliest_credible_horizon":"3_TO_12_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The candidate articulates a clinically meaningful failure hypothesis and a direct falsifier, but provides no external evidence that distributed preictal instability is present, consequential, or commonly missed by channel-wise monitoring."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"The sealed candidate identifies the single-site data custodian, investigators, ethics/IRB authority, treating epilepsy team, and patients whose consent would be required at later stages."},"distinct_testable_incremental_claim":{"status":"YES","reason":"It claims that modal gain and mode-directed pulse overlap add held-out warning and stimulation-response information beyond channel-wise signals and a nonlinear multichannel rival, with explicit null results that would defeat the claim."},"bounded_next_evidence_step":{"status":"YES","reason":"A single-site retrospective replay with existing records, patient-level splits, specified comparators, subgroup reporting, and no care changes is bounded and decision-relevant."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"For the retrospective first step, authority is identified, clinical care is unchanged, prohibited actions are explicit, and analysis claims halt when prespecified credibility or safety-related limits fail. This does not authorize prospective stimulation."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The candidate identifies major data, clinical, governance, and integration components, allowing broad resource bands, but does not specify dataset scale, staffing, software integration, regulatory pathway, trial design, or device-development requirements."}},"blocking_evidence":["No sealed evidence establishes that a stable distributed mode adds held-out value beyond channel-wise and nonlinear multichannel comparators.","The quantity, quality, accessibility, and representativeness of intracranial recordings and stimulation logs are unspecified.","Retrospective clinician-selected stimulation creates confounding that may prevent credible estimation of mode-directed intervention effects.","No evidence establishes reproducibility across patients, electrode coverage, behavioral states, preprocessing choices, or recording regimes.","Prior art is unsearched, so novelty and overlap with existing dynamical seizure-prediction or stimulation work are unknown.","Prospective safety, false-alarm burden, stimulation burden, and clinical utility remain untested."],"next_evidence_step":"Conduct the authorized single-site retrospective replay with locked preprocessing, patient-level held-out splits, and prespecified residual, spectral-gap, drift, false-alarm, and subgroup limits. Compare the modal model against channel-wise thresholds and the stated nonlinear multichannel rival for warning, then compare mode-aligned recorded pulses with matched rival-pattern, random-pattern, and no-pulse counterfactual estimates. Falsify progression if no stable distributed direction adds independent held-out value or if aligned pulses show no superior mode reduction and downstream progression metric.","research_questions":["Does a reproducible window exist in which mode identity, spectral separation, reconstruction residual, and drift all satisfy prespecified limits?","Does modal gain improve held-out warning time or accuracy without increasing false alarms relative to channel-wise thresholds and the nonlinear multichannel rival?","Does the distributed mode add information after controlling for individual-channel state, artifacts, preprocessing, and patient-level differences?","Can retrospective stimulation comparisons adequately address clinician-selection confounding, or is prospective randomized perturbation ultimately necessary?","Do recorded pulses aligned with the proposed suppressive pattern reduce the candidate mode and improve subsequent progression metrics relative to matched comparators?","How sensitive are modes and intervention maps to referencing, feature definitions, window length, electrode coverage, non-normality, and mode swapping?","Are performance and failure rates materially different across patient subgroups, seizure types, states, sites, or recording regimes?","What existing research or protected claims overlap with multivariate transition modes, preictal stability analysis, and mode-directed stimulation?"],"recommendation":"PARTNERED_RESEARCH","uncertainty_constraints":["All neuroscience efficacy and causal statements are hypotheses; no realized benefit is reported.","Problem prevalence, stakeholder demand, market size, and external distinctiveness are unmeasured.","Cost bands are resource-equivalent planning ranges, not quotations, and are highly sensitive to data readiness, regulatory scope, and device integration.","Retrospective associations cannot by themselves establish that suppressing the estimated mode prevents seizure progression.","The earliest horizon applies only to a credible retrospective result, not clinical deployment or therapeutic validation.","The assessment does not treat fitted modes as anatomical seizure sources."],"closed_book_prior_art_boundary":"Prior-art status is UNSearched. This assessment credits only the proposal's internal differentiation from its stated baseline and nearest rival; it makes no claim about novelty, field prevalence, existing trials, intellectual property, or superiority to external methods."}