{"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__gender_studies","archetype_slug":"invariant_mode_decomposition_design","domain_slug":"gender_studies","title":"Coupled-Trajectory Detection for Gender-Harm Prevention","opportunity_summary":"Test whether a bounded invariant-mode decomposition of aggregate longitudinal harm, response, and attrition measures detects escalating gender-related harm configurations earlier or more reliably than separate dashboards and prespecified intersectional longitudinal models, then use any validated signal only to formulate reversible procedural interventions.","adopter_authorizer":"An accountable institutional equity body with affected-party representation, independent privacy review, and authority limited to aggregate procedural changes.","scores":{"meaningful_impact":{"score":4,"rationale":"If the stated blind spot exists, earlier recognition of coupled escalation could improve prevention and subgroup outcomes while retaining direct harm measures; the packet provides no evidence about prevalence, effect magnitude, or how often decisions would change."},"stakeholder_pull":{"score":3,"rationale":"Affected groups, equity staff, complaint handlers, and an authorizing body are specifically identified, but the packet contains no demonstrated demand, willingness to participate, or evidence that stakeholders prioritize this analytical approach."},"incremental_advantage":{"score":3,"rationale":"The proposal makes a useful comparative claim—earlier or more reliable detection of joint escalation than separate indicators or prespecified intersectional models—but supplies no comparative results and may add little if rival models capture the same trajectories."},"distinctiveness_plausibility":{"score":2,"rationale":"The combination of modal diagnostics, direct-outcome checks, subgroup residual tests, and aggregate-use safeguards could be distinctive, but prior art is explicitly unsearched and nearby latent-state, dynamic-factor, longitudinal, and organizational-harm approaches are not compared."},"technical_implementability":{"score":3,"rationale":"The shadow analysis and comparator evaluation are technically bounded, but feasibility depends on repeated comparable measurements, adequate sample sizes, stable local dynamics, actionable spectral separation, acceptable conditioning, and privacy-preserving subgroup analysis, none of which is established."},"adoption_authority_feasibility":{"score":3,"rationale":"The packet identifies a suitably limited authority and reversible process, but affected-party representation, consent across units, independent privacy review, staff compliance with aggregate-only use, and authority over procedural changes remain unverified."},"evidence_readiness":{"score":4,"rationale":"The candidate specifies baselines, a nearest rival, held-out evaluation, direct outcomes, negative tests, and separate diagnostic and intervention falsifiers; readiness is limited by unknown data availability, comparability, subgroup coverage, and prospective evaluation duration."},"safety_net_benefit":{"score":4,"rationale":"No individual scoring is permitted, raw and qualitative evidence is retained, subgroup worsening triggers a halt, and procedures are reversible; these safeguards materially limit harm, although privacy leakage, reification, gaming, and resource diversion remain possible."},"scalability":{"score":2,"rationale":"The method cannot be deployed outside a validated regime and would require institution-specific measurement harmonization, privacy review, drift testing, stakeholder governance, and recurring validation, making transfer across organizations difficult."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"One bounded shadow-mode study using deidentified historical data and a prospectively held-out period, including data preparation, privacy review, comparator models, subgroup residual tests, qualitative discordance review, and a preregistered decision memo.","confidence":"LOW","assumptions":["An institution already holds usable longitudinal records.","No new large-scale survey or records system is required.","Small-cell privacy controls permit meaningful aggregate subgroup checks.","The study uses existing staff and contracted analytical support rather than building production software."]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Prepare one reversible procedural bundle for two consenting units with a matched or stepped comparator, including governance, consent, workflow design, secure data handling, staff training, preregistration, and evaluation setup.","confidence":"LOW","assumptions":["Shadow analysis first clears stability, residual, privacy, and incremental-performance thresholds.","The intervention changes procedures rather than adding major facilities or services.","Two units can supply an adequate comparator and evaluation window.","Independent privacy and affected-party review are available."]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Institution-wide operational launch after a successful pilot, including data-system integration, governance capacity, procedural implementation, training, independent evaluation, qualitative feedback channels, and rollback capability.","confidence":"LOW","assumptions":["The institution is medium to large and has multiple organizational units.","Existing complaint, climate, process, and retention systems require substantial integration but not full replacement.","Launch retains raw measures and survivor-defined support rather than substituting the modal metric for them.","Each new regime receives separate validation before use."]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Annual secure data operations, model and drift validation, subgroup and privacy audits, affected-party governance, staff retraining, qualitative review, direct-outcome monitoring, and independent evaluation.","confidence":"LOW","assumptions":["The model is maintained within one institution.","Recurring work includes human governance and support-process oversight, not only software maintenance.","Major redesigns or new data collection programs are excluded.","No individual-level surveillance or scoring infrastructure is operated."]}},"research_burden":"HIGH","earliest_credible_horizon":"12_TO_36_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The packet gives an observable and falsifiable account of how separate indicators could miss coupled escalation, but provides no empirical evidence that this blind spot occurs reproducibly or materially in a candidate institution."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"An accountable institutional equity body with affected-party representation, independent privacy review, and limited authority over aggregate procedural changes is explicitly identified."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The candidate can be tested against separate indicators and prespecified intersectional longitudinal models on whether it detects harmful escalation earlier and more reliably and reveals reproducible missed residual structure."},"bounded_next_evidence_step":{"status":"YES","reason":"A shadow-mode held-out analysis can be limited to deidentified aggregate data, explicit comparator models, preregistered thresholds, and rejection if stability, subgroup residual, privacy, or incremental-performance tests fail."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first step avoids live decisions and individual scoring, while the proposed governance, prohibited uses, consent requirements, and halt conditions address the principal identified safety and authority risks; their practical availability still requires partner confirmation."},"implementation_cost_scope_and_range":{"status":"YES","reason":"A bounded shadow study, two-unit pilot startup, institutional launch, and recurring operation can each be scoped into broad resource-equivalent bands, although institution size, data condition, and governance capacity make confidence low."}},"blocking_evidence":["Reproducible evidence that separate indicators and prespecified intersectional models actually miss a consequential coupled trajectory.","Availability of sufficiently repeated, comparable, privacy-preserving data with acceptable reporting, coding, participation, and attrition bias.","A stable local transition operator with usable spectral separation, conditioning, drift performance, and no unacceptable subgroup-structured residuals.","Held-out evidence that the decomposition improves detection timing or reliability over the named comparators without concealing rare severe harms.","Affected-party and privacy-review judgment that the state variables, interpretations, and proposed procedural uses are legitimate and safe.","Prior-art comparison establishing whether the analytical, governance, or falsification package offers a meaningful distinctive contribution.","For any later intervention claim, comparator-pilot evidence that the procedural bundle improves direct harm outcomes without worsening any protected subgroup outcome."],"next_evidence_step":"With one willing institutional partner and affected-party privacy oversight, preregister a retrospective-plus-held-out shadow study on a fixed deidentified aggregate dataset. Compare the proposed decomposition with separate indicator thresholds and prespecified intersectionally stratified longitudinal or event-history models on detection lead time, reliability, subgroup residuals, calibration, and direct harm outcomes. Reject advancement if it shows no reproducible incremental signal, lacks stable separation, produces subgroup-structured residuals, or conflicts materially with qualitative accounts.","research_questions":["Do coupled trajectories recur after accounting for survey participation, reporting access, coding changes, attrition, and institution-wide shocks?","Does the proposed method detect consequential escalation earlier or more reliably than separate dashboards and prespecified intersectional longitudinal models?","Are the estimated modes stable and interpretable enough for bounded procedural decisions without being reified as identities or causal structures?","Can privacy-preserving aggregation retain adequate information for small intersectional groups and rare severe harms?","Do affected parties consider the selected variables, interpretations, thresholds, and procedural responses legitimate and useful?","Which existing longitudinal, latent-state, dynamic-factor, or organizational-harm approaches already cover the claimed contribution?","Can a later reversible procedural bundle improve direct outcomes and modal gain without worsening any protected subgroup outcome?","How sensitive are conclusions to missingness, differential reporting, coding rules, temporal granularity, state-vector choices, and organizational regime changes?"],"recommendation":"PARTNERED_RESEARCH","uncertainty_constraints":["Closed-book assessment: no external validation, prevalence estimate, market evidence, or prior-art determination is available.","The problem is plausible and testable but not shown to occur materially in any organization.","All cost bands are low-confidence resource-equivalent scenarios because institution size, data readiness, compliance requirements, and evaluation duration are unspecified.","Descriptive modal structure cannot establish causation or be treated as a natural taxonomy of gendered experience.","Reporting, participation, coding, attrition, reflexivity, and institutional shocks may generate apparent co-movement.","Any evidence is local to the validated organizational regime; transfer and scale require renewed validation.","The earliest credible horizon assumes an existing partner with usable longitudinal data and functioning affected-party and privacy governance."],"closed_book_prior_art_boundary":"Prior-art status is explicitly UNSEARCHED. No conclusion is made about novelty, prevalence, market size, realized impact, or superiority over intersectional longitudinal, multilevel event-history, latent-state, dynamic-factor, or organizational-harm monitoring approaches."}