{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"invariant_mode_decomposition_design__speech_language_pathology","trajectory_id":"R","attempt_index":0,"archetype_slug":"invariant_mode_decomposition_design","domain_slug":"speech_language_pathology","decision":"CANDIDATE","problem_id":"dysarthria_coupled_subsystem_targeting_failure","causal_lever_id":"reproducible_multivariate_speech_response_modes","proposal":{"problem":"In dysarthria rehabilitation, respiratory, phonatory, articulatory, prosodic, and rate measures may change together, while treatment is commonly organized around individually visible impairments. When intelligibility depends on reproducible combinations rather than single measures, clinicians can select low-leverage targets, miss a worsening coupled pattern, or mistake task-specific residual behavior for a stable treatment target. Whether such modes exist and remain stable is a HYPOTHESIS.","actors_substrate":["Adults with stable acquired dysarthria who consent to repeated speech probes","Treating speech-language pathologists","Communication partners affected by intelligibility and participation outcomes","Repeated recordings from standardized speaking tasks","Multivariate acoustic, perceptual, respiratory, phonatory, articulatory, rate, and intelligibility measures"],"observable_state":"For each task and probe, a standardized vector records subsystem measures plus blinded intelligibility and effort outcomes; repeated observations estimate a local transition from pre-probe state to post-probe state. Mode coefficients, gains, reconstruction residuals, and between-session drift are observable analysis outputs.","consequence":"If a reproducible weakly damped or high-leverage combination is hidden across named subsystems, coordinate-by-coordinate treatment may spend effort on salient variables without improving intelligibility and may fail to detect deterioration until surface outcomes worsen.","affected_objective":"Improve intelligibility and communicative participation per unit treatment burden while preserving comfort, autonomy, and clinical safety.","structural_mapping":[{"archetype_element":"Coupled transformation","domain_realization":"A locally estimated mapping from a multivariate speech state before a standardized rate, loudness, or articulatory probe to the state and intelligibility response afterward.","claim_kind":"HYPOTHESIS"},{"archetype_element":"Invariant directions","domain_realization":"Approximately repeatable combinations of subsystem changes that preserve their direction across probe repetitions within a declared task and severity window.","claim_kind":"HYPOTHESIS"},{"archetype_element":"Scalar modal response","domain_realization":"Estimated persistence or amplification of each coupled response pattern, supplemented by its sensitivity to intelligibility and effort.","claim_kind":"INFERENCE"},{"archetype_element":"Action-relevant modes","domain_realization":"Reproducible modes whose perturbation predicts clinically meaningful outcome change without exceeding burden or safety limits.","claim_kind":"HYPOTHESIS"},{"archetype_element":"Residual and drift governance","domain_realization":"Held-out reconstruction error, structured residual inspection, mode-angle drift, and spectral-gap checks determine whether the model remains usable.","claim_kind":"INFERENCE"}],"component_map":[{"component":"Transformation Scope","status":"adapted","domain_realization":"Within-person, task-specific pre/post response operator under standardized low-risk speech probes."},{"component":"State-Vector Definition","status":"direct","domain_realization":"Predeclared standardized subsystem, intelligibility, and effort variables with fixed units and missing-data rules."},{"component":"Invariant Mode Basis","status":"adapted","domain_realization":"Approximate eigenvectors of the estimated local response operator, interpreted only as coupled response patterns."},{"component":"Modal Gain Spectrum","status":"direct","domain_realization":"Eigenvalues estimate decay, persistence, or growth within the sampled window."},{"component":"Dominant Mode Selection Rule","status":"direct","domain_realization":"Retain modes meeting reproducibility, outcome-sensitivity, and held-out residual thresholds, not magnitude alone."},{"component":"Stable/Unstable Mode Partition","status":"adapted","domain_realization":"Classify modes as damped, persistent, oscillatory, or growing only across the repeated probe horizon."},{"component":"Modal Intervention Map","status":"adapted","domain_realization":"Map feasible clinician-controlled probes to predicted changes in mode coordinates and outcomes."},{"component":"Reconstruction Residual Check","status":"direct","domain_realization":"Reconstruct held-out original-variable responses and inspect both error magnitude and clinically structured omissions."},{"component":"Mode Drift Monitor","status":"direct","domain_realization":"Compare mode directions, ordering, and gains across sessions and tasks."},{"component":"Interpretation Scope Contract","status":"direct","domain_realization":"State that modes are local descriptive response patterns, not anatomical modules or globally causal disease mechanisms."},{"component":"Mode-Coupling Register","status":"direct","domain_realization":"Record cross-mode movement, near-degenerate modes, and non-orthogonal sensitivity effects."},{"component":"Local Linearization Window","status":"adapted","domain_realization":"Specified diagnosis stability, task, cue intensity, fatigue range, and session interval."},{"component":"Spectral Gap Threshold","status":"direct","domain_realization":"Predeclared minimum separation and bootstrap stability required to treat retained modes as distinct."}],"mechanism_dispositions":[{"slug":"eigendecomposition_workflow","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Decompose the explicitly estimated local response operator; report conditioning and complex or incomplete bases.","counterfactual_removal":"Without it, the proposed invariant directions and modal gains are not obtained."},{"slug":"modal_sensitivity_sweep","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Use bounded clinician-feasible probes and measure intelligibility, effort, and cross-mode responses.","counterfactual_removal":"Mode magnitude could be mistaken for clinical leverage, defeating target selection."},{"slug":"modal_stability_analysis","disposition":"selected_supporting","contribution_type":"TEST_DESIGN","adaptation_or_rejection":"Classify short-horizon repeated-probe behavior only within the declared local window.","counterfactual_removal":"The study could rank leverage but could not test whether a concerning pattern grows or dissipates."},{"slug":"mode_shape_testing","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Adapt excitation to standardized, consented speech probes and infer empirical response shapes from recordings.","counterfactual_removal":"Analytical modes would lack a direct check against observed patient responses."},{"slug":"network_spectral_centrality_analysis","disposition":"incompatible","contribution_type":"NONE","adaptation_or_rejection":"The proposal concerns coupled measured variables, not a justified node-connectivity operator or node-importance question.","counterfactual_removal":"No change; node centrality is not in the causal chain."},{"slug":"power_iteration_probe","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"The pilot uses a small operator and needs multiple modes and residual reconstruction; dominant-only iteration is unnecessary and unreliable near a small gap.","counterfactual_removal":"No material change because full decomposition supplies the required spectrum."},{"slug":"principal_component_analysis","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Variance directions do not establish response dynamics or intervention leverage; PCA may be reported only as a noncausal rival.","counterfactual_removal":"No change to the transformation-based causal chain."},{"slug":"reduced_order_model","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"A runnable surrogate is premature before reproducibility and prospective leverage are established.","counterfactual_removal":"The bounded test can score candidate probes directly without simulation."},{"slug":"residual_reconstruction_test","disposition":"selected_load_bearing","contribution_type":"TEST_DESIGN","adaptation_or_rejection":"Select mode count using held-out error plus structured clinical residuals.","counterfactual_removal":"Compression could silently discard low-energy behavior important to intelligibility or safety."},{"slug":"singular_value_decomposition","disposition":"selected_supporting","contribution_type":"TEST_DESIGN","adaptation_or_rejection":"Use singular values and conditioning diagnostically when the estimated operator is non-normal; do not relabel singular vectors as invariant eigenmodes.","counterfactual_removal":"Ill-conditioned eigenmode claims would be less readily detected, though the primary analysis remains possible."},{"slug":"spectral_decomposition_report","disposition":"selected_supporting","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Provide clinicians a bounded interpretation, uncertainty statement, coupling register, and prohibited readings.","counterfactual_removal":"Tentative mathematical patterns could be misread as anatomical facts or treatment mandates."},{"slug":"spectral_gap_monitor","disposition":"selected_supporting","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Re-estimate gaps and mode alignment at each study session; suspend use after threshold crossing.","counterfactual_removal":"A fixed basis could continue guiding probes after mode identity becomes unstable."}],"causal_chain":["Standardized repeated probes expose multivariate pre/post speech responses.","A locally valid operator summarizes how combinations of measures change together.","Decomposition identifies approximately preserved directions and their short-horizon gains.","Sensitivity testing separates outcome-relevant modes from merely large modes.","Residual, conditioning, gap, and drift checks gate whether the modal interpretation is usable.","Within that gate, clinicians test a feasible probe aimed at the highest-leverage mode.","Prospective intelligibility and burden outcomes determine whether modal targeting adds value."],"baseline":"Ordinary impairment-oriented care: the SLP integrates perceptual assessment and named subsystem measures, then selects and adjusts targets from salient deficits and observed response without an explicit multivariate transition model.","nearest_rival":"A preregistered coordinate-level response model that ranks individual speech variables by their prospective association with intelligibility, using the same observations and complexity budget but no modal decomposition.","authority_safety":{"affected_parties":["Participating patients","Communication partners","Treating and evaluating clinicians"],"decision_authority":"The patient retains consent and withdrawal authority; the licensed treating SLP retains clinical authority; an approved study protocol governs data use and probe limits.","authorized_first_step":"Run a bounded feasibility study in consenting adults with clinically stable dysarthria: collect repeated low-risk standardized probes, fit models on early repetitions, and prospectively compare one mode-guided probe with the coordinate-level rival using blinded intelligibility, effort, discomfort, and held-out residuals. No permanent care change follows from this test.","excluded_actions":["Withholding indicated care","Automatically changing treatment from model output","Interpreting modes as lesions or independent physiological modules","Using unvalidated modes across diagnoses, tasks, or severity regimes","Applying probes that create respiratory, vocal, swallowing, fatigue, or distress risk"],"halt_rollback":"Stop a probe for patient request, discomfort, fatigue, adverse voice or breathing signs, residual above tolerance, poor conditioning, gap below threshold, or excessive mode drift. Return immediately to clinician-directed care and retire the model until reviewed and re-estimated."}},"negative_tests":{"strongest_counterevidence":"Speech responses may be strongly nonlinear, voluntary, task-dependent, and nonstationary; limited repeated observations may yield unstable eigenvectors, while ordinary clinician-selected single targets may already capture most actionable variance.","analogy_break":"Mathematical modes need not be independent biological control units. Unlike a fixed physical operator, a speaker adapts strategically, and fatigue, learning, cue interpretation, and context can change the transformation during measurement.","failure_condition":"No mode set simultaneously achieves preregistered reproducibility, acceptable conditioning, spectral separation, held-out reconstruction, and a usable task window.","problem_falsifier":"Across repeated tasks, individual-variable models are as reproducible and prospectively predictive as combination models, perturbations show negligible cross-variable structure, and no hidden coupled pattern adds clinically relevant explanation.","intervention_falsifier":"Reproducible modes are found, but mode-guided probes do not improve blinded intelligibility or participation-proximal outcomes over the coordinate-level rival, or they increase effort, discomfort, or variability.","risks":["Overfitting a small repeated-measures dataset","Confusing descriptive modes with causal physiology","Mode swapping or instability near a small spectral gap","Suppressing clinically important low-energy residual behavior","Extrapolating across tasks, fatigue states, diagnoses, or patients","Added assessment burden and recording privacy exposure"]},"null_rationale":null,"classification":{"candidate_kind":"MECHANISM_ADAPTATION","prior_art_status":"UNSEARCHED","evidence_maturity":"HYPOTHESIS"},"revision_change_log":{"revision_kind":"ORIGINAL","prior_problem_id":null,"prior_causal_lever_id":null,"problem_changed":false,"causal_lever_changed":false,"conceptual_changes":[],"operational_changes":[],"repairs_addressed":[]},"confidence":0.72,"generator_notes":"Closed-book structural inference from the supplied packet. Domain-specific prevalence, effect size, feasibility, and novelty were not established; empirical statements are therefore framed as hypotheses or bounded inferences."}