{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp04_retrieval_first_paired20_20260802","cell_id":"invariant_mode_decomposition_design__psychology","arm":"RETRIEVAL_FIRST","round_index":0,"hypotheses":[{"hypothesis_id":"H1","title":"Unstable-mode targeting in individualized CBT","problem":"Clinicians target the largest reported symptom even when a coupled symptom-behavior pattern is driving deterioration.","affected_stakeholder":"Adults receiving CBT with repeated ecological assessments","workflow_boundary":"Between weekly measurement review and selection of the next CBT module","failure_mode":"Coordinate-level symptom targeting misses a smaller but growing avoidance-affect mode.","unit_of_analysis":"Within-patient daily transition among affect, avoidance, sleep, rumination, and functioning","causal_lever":"Assign the next CBT module to the locally unstable mode with the greatest outcome sensitivity.","archetype_mapping":"Estimate the patient transition operator, partition modes by stability, perturb candidate module effects in modal coordinates, and retain residual checks.","expected_value":"Potentially faster functional improvement with fewer low-leverage module changes.","falsifiable_claim":"In an 8-week randomized pilot, mode-guided module selection will improve functioning at least 0.3 standardized units more than largest-symptom targeting; failure to reach that difference rejects the claim.","diversity_rationale":"Targets treatment-module selection for one patient's dynamic state, unlike pacing, group influence, intake compression, or relapse monitoring.","mechanism_slugs":["eigendecomposition_workflow","modal_stability_analysis","modal_sensitivity_sweep","residual_reconstruction_test"],"search_questions":["Have psychotherapy systems selected modules from eigenmodes of within-person symptom transitions?","Do unstable within-person modes predict functional decline beyond symptom levels?","Can CBT module effects be estimated reliably in modal coordinates?"]},{"hypothesis_id":"H2","title":"Mode-adaptive exposure pacing","problem":"Exposure hierarchies advance on mean fear reduction although approach, avoidance, arousal, and safety behavior may jointly oscillate or remain marginally stable.","affected_stakeholder":"Patients receiving exposure therapy for anxiety disorders","workflow_boundary":"Between completion of one exposure trial and difficulty selection for the next trial","failure_mode":"A scalar fear rule advances during unstable avoidance dynamics or stalls during harmless arousal persistence.","unit_of_analysis":"Trial-to-trial transition of a patient's multivariate exposure response","causal_lever":"Increase, hold, or reduce exposure difficulty according to the stability class and sensitivity of the dominant response mode.","archetype_mapping":"Empirically excite the response system with graded trials, recover mode shapes, classify their dynamics, and bound interpretation to the tested exposure range.","expected_value":"Potentially fewer premature escalations and fewer unnecessary repetitions.","falsifiable_claim":"Across four sessions, mode-adaptive pacing will reduce safety behavior at least 20% more than fear-score pacing without increasing dropout by over 5 percentage points.","diversity_rationale":"Changes moment-to-moment dose pacing using experimentally excited trial dynamics rather than selecting treatment modules or monitoring longer-term care.","mechanism_slugs":["mode_shape_testing","modal_stability_analysis","modal_sensitivity_sweep"],"search_questions":["Has exposure difficulty been adapted using multivariate dynamical modes rather than fear thresholds?","Which trial-level variables yield stable and identifiable exposure modes?","Does modal instability predict dropout or symptom rebound?"]},{"hypothesis_id":"H3","title":"Spectral facilitation in group therapy","problem":"Facilitators distribute attention by speaking time or visible distress, missing members who structurally sustain avoidance or engagement norms.","affected_stakeholder":"Participants and facilitators in group psychotherapy","workflow_boundary":"During facilitator choice of whom to invite, pair, or redirect in a session","failure_mode":"Local prominence measures misidentify the members through whom behavioral patterns propagate.","unit_of_analysis":"Directed member-to-member influence network within a therapy group","causal_lever":"Direct facilitator prompts or peer-modeling roles toward members with high centrality in the maladaptive or therapeutic dominant mode.","archetype_mapping":"Treat observed interpersonal influence as a network operator, extract its dominant eigenvector, and translate node loadings into bounded facilitation actions.","expected_value":"Potentially stronger diffusion of therapeutic participation from the same facilitator time.","falsifiable_claim":"In randomized group sessions, spectral targeting will increase next-session therapeutic participation by at least 15% relative to speaking-time targeting; a smaller difference rejects the claim.","diversity_rationale":"Uses a group-level network and facilitator allocation lever, structurally distinct from individual state dynamics and assessment design.","mechanism_slugs":["network_spectral_centrality_analysis","power_iteration_probe","spectral_decomposition_report"],"search_questions":["Has eigenvector centrality guided real-time psychotherapy-group facilitation?","Can directed influence networks be estimated reliably from session behavior?","Does targeting central members alter group norms more than targeting frequent speakers?"]},{"hypothesis_id":"H4","title":"Treatment-preserving compressed intake","problem":"Short intake batteries discard low-variance combinations of responses that may distinguish which treatment will work.","affected_stakeholder":"Patients entering high-volume outpatient mental-health services","workflow_boundary":"Between intake assessment and assignment to a treatment pathway","failure_mode":"Variance-based item reduction preserves common severity while erasing treatment-response moderators.","unit_of_analysis":"Patient-by-intake-feature mapping to treatment-specific outcomes","causal_lever":"Retain items spanning singular directions with high differential-treatment sensitivity, including structured residual directions.","archetype_mapping":"Factor the rectangular intake-to-outcome mapping, construct a reduced assessment from action-relevant modes, and test out-of-sample reconstruction and decision residuals.","expected_value":"Potentially shorter intake without degrading treatment matching.","falsifiable_claim":"A reduced battery using at most half the items will be noninferior to the full battery within 3 percentage points for selecting the better of two treatments on 12-week outcome.","diversity_rationale":"Redesigns a measurement artifact at population intake, rather than intervening on therapy dynamics or social influence.","mechanism_slugs":["singular_value_decomposition","reduced_order_model","residual_reconstruction_test","modal_sensitivity_sweep"],"search_questions":["Have mental-health intake batteries been compressed against differential treatment response rather than variance?","Do low-variance residual directions contain reproducible treatment moderators?","What prospective noninferiority margins are acceptable for treatment-matching assessments?"]},{"hypothesis_id":"H5","title":"Modal-drift relapse escalation","problem":"Maintenance care escalates only after total symptom scores cross a threshold, delaying response when the patient's relapse pattern has changed.","affected_stakeholder":"Recently discharged patients receiving relapse-prevention monitoring","workflow_boundary":"Between remote weekly check-in and clinician outreach or booster treatment","failure_mode":"Stable aggregate scores conceal rotation, amplification, or gap collapse among coupled relapse indicators.","unit_of_analysis":"Rolling within-patient weekly spectrum of sleep, affect, activity, cognition, and social withdrawal","causal_lever":"Trigger reassessment when dominant modes drift, the retained-mode spectral gap collapses, or structured residuals grow.","archetype_mapping":"Run a reduced modal monitor within its validated window and retire or refit it when gap, direction, or residual thresholds fail.","expected_value":"Potentially earlier, more specific booster care with fewer symptom-threshold alerts.","falsifiable_claim":"Over six months, modal-drift alerts will identify relapse at least two weeks earlier than total-score thresholds while keeping false alerts below one per patient-month.","diversity_rationale":"Uses longitudinal model-validity signals to trigger post-discharge escalation, distinct from treatment choice, session pacing, group intervention, and intake compression.","mechanism_slugs":["spectral_gap_monitor","principal_component_analysis","residual_reconstruction_test","spectral_decomposition_report"],"search_questions":["Have spectral-gap or mode-rotation alarms been tested for psychiatric relapse monitoring?","How stable are within-patient relapse modes across recovery regimes?","Do modal-drift alerts add lead time beyond symptom slopes and threshold rules?"]}]}