{"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__psychology","archetype_slug":"invariant_mode_decomposition_design","domain_slug":"psychology","title":"Safety-gated person-specific symptom-mode monitoring for recurrent depression","opportunity_summary":"Test whether stable, person-specific combinations of rumination, withdrawal, sleep disruption, and negative affect can provide earlier or more actionable warning of depressive deterioration than total scores, univariate thresholds, or an unrestricted multivariate forecast. The proposed value is plausible but remains contingent on mode stability, incremental prediction, acceptable burden, and eventual causal actionability.","adopter_authorizer":"An outpatient mental-health service could adopt the workflow; participating patients authorize data collection, licensed treating clinicians retain all care and escalation authority, and the service's research, privacy, and clinical-governance bodies would authorize evaluation and deployment.","scores":{"meaningful_impact":{"score":4,"rationale":"If the hypothesized coupled escalation exists, earlier review and better targeting could reduce missed deterioration and misdirected intervention. The packet does not establish how often this failure occurs or whether acting on a mode improves outcomes, preventing a very favorable score."},"stakeholder_pull":{"score":3,"rationale":"Patients and clinicians have a plausible interest in safer early warning during recurrent-depression recovery, but the packet contains no evidence of demand, willingness to sustain intensive measurement, tolerance for alerts, or workflow fit."},"incremental_advantage":{"score":3,"rationale":"The candidate makes a concrete comparison against symptom-by-symptom review and a regularized full multivariate forecast, with possible advantages in lead time and intervention ranking. No held-out result yet shows improvement over either comparator."},"distinctiveness_plausibility":{"score":2,"rationale":"Invariant-mode reduction, stability gates, and mode-mapped intervention targeting differentiate the proposal internally from the stated nearest rival, but prior art is explicitly unsearched and distinctiveness from existing personalized dynamical symptom models is unknown."},"technical_implementability":{"score":3,"rationale":"Repeated measurements, local transition fitting, held-out evaluation, and drift monitoring are technically specifiable. Implementability is constrained by noisy and missing psychological measurements, person-level data requirements, non-normal or poorly conditioned operators, mode rotation, and changing dynamics during treatment."},"adoption_authority_feasibility":{"score":4,"rationale":"The packet clearly assigns participation and data authority to the patient and clinical authority to the licensed clinician, while prohibiting autonomous diagnosis, treatment, and crisis action. Institutional privacy, compliance, and workflow authorization remain untested."},"evidence_readiness":{"score":4,"rationale":"A silent prospective pilot is explicitly authorized and includes preregistered comparisons, held-out prediction, calibration, lead time, residuals, and false-alert burden, together with strong falsifiers and halt rules. Protocol parameters and data sufficiency are not yet established."},"safety_net_benefit":{"score":4,"rationale":"Usual care remains in place, outputs can be suppressed, existing escalation protocols govern action, and the first study produces no patient-facing alerts. Residual risks include false reassurance, privacy loss, symptom hypervigilance, and later clinician overinterpretation."},"scalability":{"score":2,"rationale":"The approach requires consented intensive longitudinal measurement, person-specific model fitting, stability and drift checks, and clinician-mediated interpretation. The packet provides no evidence that data density, monitoring labor, model portability, or false-alert burden will remain acceptable across patients and sites."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"250K_TO_1M","scope":"One-site, time-bounded prospective silent-mode study using consented historical-plus-new measurements, including protocol development, participant coordination, secure data handling, model development, comparator evaluation, clinical oversight, and analysis.","confidence":"LOW","assumptions":["An outpatient clinical partner and some existing measurement infrastructure are available.","No patient-facing alerts or experimental treatment changes occur.","The cohort and observation window are sufficient for preliminary stability and held-out comparisons but not definitive clinical-effectiveness evidence."]},"initial_deployment_startup":{"band_2026_usd":"1M_TO_5M","scope":"Clinical-grade preparation for limited deployment within one health system, including data pipelines, software hardening, security and privacy review, clinical validation, workflow integration, governance, training, and monitoring controls.","confidence":"LOW","assumptions":["Deployment follows successful silent validation.","The system remains clinician-facing and does not exercise autonomous care authority.","Integration must accommodate passive and self-report data, missingness, drift, auditability, and output suppression."]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Limited launch in one outpatient program with clinician review, patient consent operations, support, quality monitoring, incident response, and prospective evaluation against usual review.","confidence":"LOW","assumptions":["Core software and governance work are completed during startup.","Launch is restricted to a bounded patient population and trained clinical team.","Usual care and existing crisis protocols remain fully active."]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Annual operation for one health-system program, covering data collection, hosting, security, model refitting and drift review, clinician and participant support, compliance, maintenance, and continuing evaluation.","confidence":"LOW","assumptions":["Measurement and clinician-review intensity remain substantial.","No specialized hardware is required beyond participant devices and existing clinical infrastructure.","Costs could rise materially with multiple sites, larger populations, or continuous passive-data integration."]}},"research_burden":"HIGH","earliest_credible_horizon":"3_TO_12_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The candidate specifies an observable and falsifiable failure of symptom-by-symptom monitoring, but provides no external evidence that missed coupled escalation occurs reproducibly or often enough to justify adoption."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"The packet identifies participating patients, licensed treating clinicians, and outpatient care settings, and explicitly assigns consent and clinical decision authority."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The proposal can be tested for improved held-out prediction, calibration, lead time, residual behavior, false-alert burden, and intervention ranking relative to usual univariate review and a regularized full multivariate forecast."},"bounded_next_evidence_step":{"status":"YES","reason":"The authorized silent-mode pilot is prospective, consented, preregistered, comparator-based, non-interventional, and subject to explicit suppression and halt criteria."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first step preserves usual care, issues no patient-facing alerts, prohibits autonomous clinical decisions, retains clinician authority, limits data to explicit consent, and defines rollback conditions."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The candidate identifies required measurements, modeling, clinical coordination, and safeguards but supplies no cohort size, duration, site count, integration environment, staffing model, or resource estimate; the cost bands therefore depend on explicit evaluator assumptions."}},"blocking_evidence":["Prospective evidence that reproducible coupled deterioration patterns exist beyond directly observed symptoms or total-score change.","Evidence that estimated modes retain stable identity, acceptable conditioning, spectral separation, low structured residuals, and useful outcome sensitivity across held-out windows.","Evidence of incremental accuracy, calibration, or clinically useful lead time over both usual symptom monitoring and the full multivariate forecasting rival.","Evidence that measurement burden, missingness, privacy requirements, and false-alert rates are acceptable to patients and clinicians.","Clinician-controlled evidence that mode-mapped support improves mode amplitude or deterioration outcomes relative to an intensity-matched symptom-targeted intervention.","External prior-art evidence establishing whether the modal formulation or safety-gated workflow is meaningfully distinct."],"next_evidence_step":"Run the authorized one-site, time-bounded silent prospective study with preregistered variables, thresholds, stability gates, and evaluation windows. Compare the modal model with total-score and univariate monitoring and with a regularized full multivariate forecast on held-out accuracy, calibration, lead time, residual structure, and false-alert burden. Stop advancement if direct indicators perform as well, the multivariate rival matches or exceeds performance, modes rotate across resamples, or conditioning, spectral separation, residual, missingness, or burden limits fail.","research_questions":["Do person-specific amplifying modes remain identifiable and stable across resamples and clinically relevant time windows?","Does modal monitoring add calibrated lead time or reduce missed deterioration relative to total scores, univariate thresholds, and a full multivariate forecast?","How much within-person measurement density is required, and how sensitive are outputs to missingness, treatment changes, reporting reactivity, and measurement frequency?","Can contributors to a stable mode be mapped to clinician-approved actions without treating the mode as a diagnosis or biological entity?","Does mode-mapped support outperform an intensity-matched symptom-targeted rival under clinician-controlled assignment?","What false-alert burden, explanation format, consent model, and privacy controls are acceptable to patients, clinicians, and institutional authorizers?","Is the modal formulation or its safety-gated workflow distinct from existing personalized dynamical symptom modeling?"] ,"recommendation":"PARTNERED_RESEARCH","uncertainty_constraints":["Problem prevalence and the frequency of missed multimodal escalation are unsupported in the sealed packet.","The existence, stability, conditioning, spectral separation, and clinical usefulness of person-specific modes are hypotheses.","Predictive improvement over the baseline and nearest rival has not been demonstrated.","Model perturbation supports only causal plausibility, not identified intervention effects.","Patient and clinician demand, burden tolerance, workflow fit, and institutional approval are unmeasured.","Deployment scale, integration requirements, staffing, and exact costs are unspecified.","External novelty, prior-art position, market size, and realized impact cannot be assessed closed-book."],"closed_book_prior_art_boundary":"No external search was performed. The sealed candidate is internally differentiated from its stated full-multivariate rival, but prior-art status is explicitly UNSEARCHED; no claim is made about novelty, prevalence, market position, or existing implementations."}