{"schema_version":1,"research_id":"eoa_inverse_innovation_exp04_external_evaluation_20260802","source_assessment_id":"invariant_mode_decomposition_design__medicine_healthcare:DISCARDED_AUDIT:H2:v0","cell_id":"invariant_mode_decomposition_design__medicine_healthcare","search_queries":["intradialytic hypotension prevalence symptomatic hemodialysis systematic review primary research","KDIGO intradialytic hypotension ultrafiltration rate blood volume monitoring guidance","randomized trial biofeedback hemodialysis blood volume ultrafiltration intradialytic hypotension","patient-specific mathematical model hemodialysis cardiovascular response ultrafiltration control trial","real-time prediction intradialytic hypotension machine learning cloud computing","novel ultrafiltration rate feedback controller hemodialysis clinical pilot","computer control regulation blood volume heart rate blood pressure kidney dialysis","hemodialysis SVD singular value decomposition ultrafiltration blood pressure patient-specific model"],"sources":[{"source_id":"S1","title":"The Prevalence of Intradialytic Hypotension in Patients on Conventional Hemodialysis: A Systematic Review with Meta-Analysis","publisher":"American Journal of Nephrology (Karger)","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC6604263/","source_class":"AUTHORITATIVE_SECONDARY","publication_date":"2019-05-24","accessed_at":"2026-08-02","claims_supported":["IDH is frequent, clinically consequential, and definition-dependent.","Meta-analysis estimated IDH in 10.1% of sessions under the EBPG definition and 11.6% under a nadir-SBP-below-90 definition.","Diabetes, higher interdialytic weight gain, female sex, and lower body weight were recurrent risk factors."]},{"source_id":"S2","title":"Blood pressure and volume management in dialysis: conclusions from a Kidney Disease: Improving Global Outcomes (KDIGO) Controversies Conference","publisher":"Kidney Disease: Improving Global Outcomes / Kidney International","url":"https://kdigo.org/wp-content/uploads/2017/05/KDIGO-BP-Volume-in-Dialysis-FINAL.pdf","source_class":"OFFICIAL_GUIDANCE","publication_date":"2020-03-08","accessed_at":"2026-08-02","claims_supported":["IDH reflects both inadequate intravascular volume for the desired ultrafiltration rate and inadequate cardiovascular compensation.","Symptomatic BP decreases or nadir systolic BP below 90 mmHg should trigger reassessment of ultrafiltration rate, treatment time, interdialytic weight gain, target weight, and medications.","Dialysis prescription decisions occupy a narrow therapeutic window because preventing hypotension must not compromise euvolemia or adequate treatment.","Individualization should incorporate hemodynamics, comorbidities, symptoms, clinical judgment, preferences, and local resources."]},{"source_id":"S3","title":"Randomized Crossover Trial of Blood Volume Monitoring–Guided Ultrafiltration Biofeedback to Reduce Intradialytic Hypotensive Episodes with Hemodialysis","publisher":"Clinical Journal of the American Society of Nephrology","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC5672962/","source_class":"PRIMARY_RESEARCH","publication_date":"2017-10-10","accessed_at":"2026-08-02","claims_supported":["Automated ultrafiltration adjustment based on blood-volume parameters is clinically implementable.","In 32 randomized hypotension-prone patients, blood-volume-guided ultrafiltration biofeedback did not significantly reduce symptomatic IDH relative to best clinical practice.","Existing biofeedback evidence is heterogeneous and does not establish that control from a single monitored coordinate improves outcomes."]},{"source_id":"S4","title":"Blood volume controlled hemodialysis in hypotension-prone patients: a randomized, multicenter controlled trial","publisher":"Kidney International","url":"https://pubmed.ncbi.nlm.nih.gov/12164888/","source_class":"PRIMARY_RESEARCH","publication_date":"2002-09-01","accessed_at":"2026-08-02","claims_supported":["A clinical system already continuously modified ultrafiltration and dialysate conductivity to track a prescribed blood-volume trajectory.","In a 36-patient, 10-center randomized crossover trial, the system reduced IDH events by about 30% relative to conventional dialysis.","The proposed 20% reduction target lies within an effect range previously reported for related closed-loop technology."]},{"source_id":"S5","title":"Can the response to dialysis treatment be predicted by using patient-specific modeling of fluid and solute exchanges? A multicentric evaluation","publisher":"Artificial Organs (Wiley)","url":"https://onlinelibrary.wiley.com/doi/full/10.1111/aor.14530","source_class":"PRIMARY_RESEARCH","publication_date":"2023-04-10","accessed_at":"2026-08-02","claims_supported":["Patient-specific dialysis-response models can be fitted from three prior sessions and evaluated on three later sessions.","Across 68 patients, average normalized predictive error increased by only 0.97 percentage points between training and independent sessions.","The study predicted solute concentrations and hematic-volume change but did not validate a setting-selection policy against symptomatic IDH or Kt/V outcomes."]},{"source_id":"S6","title":"Real-time prediction of intradialytic hypotension using machine learning and cloud computing infrastructure","publisher":"Nephrology Dialysis Transplantation (Oxford University Press)","url":"https://pubmed.ncbi.nlm.nih.gov/37055366/","source_class":"PRIMARY_RESEARCH","publication_date":"2023-06-30","accessed_at":"2026-08-02","claims_supported":["Real-time IDH prediction using machine and clinical data is technically feasible.","A model trained on 42,656 sessions from 693 patients achieved AUROC 0.89 for prediction 15–75 minutes ahead.","Recent systolic BP and outcomes from the previous ten sessions were important predictors, supporting repeated-session personalization while leaving intervention benefit untested."]},{"source_id":"S7","title":"A Novel Ultrafiltration Rate Feedback Controller for Use in Hemodialysis: First Clinical Experience: An Interventional Pilot Study","publisher":"Kidney360 (American Society of Nephrology)","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC12483043/","source_class":"PRIMARY_RESEARCH","publication_date":"2025-06-04","accessed_at":"2026-08-02","claims_supported":["A bounded, clinician-reviewed ultrafiltration recommendation controller was prospectively operated during 63 treatments in 15 patients.","The controller issued 1,037 recommendations and increased relative-blood-volume target attainment from 47% to 69%.","Mean BP did not improve and nadir systolic BP was lower with the controller, illustrating that better trajectory control does not necessarily improve hypotension outcomes.","Licensed staff review, prescription-compatible bounds, fallback conditions, and hard stops were used as safety controls."]},{"source_id":"S8","title":"A computer control system for the regulation of blood volume, heart rate and blood pressure during kidney dialysis","publisher":"IFAC Proceedings Volumes (Elsevier)","url":"https://www.sciencedirect.com/science/article/pii/S1474667016459215","source_class":"PRIMARY_RESEARCH","publication_date":"2011-08-28","accessed_at":"2026-08-02","claims_supported":["A multivariable dialysis controller already used ultrafiltration rate and dialysate sodium as inputs and relative blood volume, heart-rate change, and systolic BP as outputs.","The system used a linear parameter-varying response model, constrained model-predictive control, and an optimization objective.","The controller was experimentally evaluated in patients, creating close functional prior art to model-based selection of settings intended to damp hypotension-sensitive physiology."]}],"problem_evidence":{"support":"STRONG","rationale":"Systematic-review and KDIGO evidence establishes a recurrent, consequential IDH problem arising from coupled volume removal and cardiovascular compensation. The literature also supports the hypothesis's criticism of BP-only adjustment: clinically relevant decisions must consider ultrafiltration, target weight, treatment time, symptoms, comorbidities, and compensatory physiology together.","source_ids":["S1","S2"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"Official guidance prioritizes individualized management, and multiple multicenter trials enrolled hypotension-prone patients and required clinician or nursing interaction with automated controls. This demonstrates a recognized clinical need and workflow relevance, but no source shows a dialysis organization, nephrologist group, patient group, or regulator requesting the particular SVD-based method as written.","source_ids":["S2","S3","S4","S7"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"Constrained multivariable model-predictive dialysis control","similarity":"Very high functional similarity: a coupled hemodynamic response model maps ultrafiltration and dialysate sodium to blood-volume, heart-rate, and BP outcomes, then selects bounded controls through optimization.","remaining_difference":"It uses LPV/MPC rather than SVD-derived modes, a modal sensitivity sweep, and an explicit out-of-sample reconstruction gate; the reported endpoint was not the proposed symptomatic-IDH/Kt/V composite.","source_ids":["S8"]},{"name":"Blood-volume tracking with automated ultrafiltration and conductivity adjustment","similarity":"Established clinical biofeedback already varies the same principal machine settings to follow a patient-relevant physiological trajectory and has been tested against IDH.","remaining_difference":"It follows a prescribed blood-volume trajectory rather than decomposing a patient-specific multivariable setting-to-physiology response map into outcome-sensitive modes.","source_ids":["S3","S4"]},{"name":"Patient-specific multipool dialysis-response modeling","similarity":"It learns patient-specific coupled fluid and solute responses from repeated sessions and tests predictions on later sessions, closely matching the proposed longitudinal estimation and residual-validation premise.","remaining_difference":"It predicts trajectories but does not rank feasible controls in modal coordinates or test whether its recommendations reduce symptomatic IDH while preserving Kt/V.","source_ids":["S5"]},{"name":"Clinician-reviewed ultrafiltration feedback controller","similarity":"It produces frequent, bounded ultrafiltration recommendations from live physiology within an actual dialysis workflow and includes human review and safety stops.","remaining_difference":"It is a proportional-integral target-trajectory controller, not an SVD/modal intervention policy, and it did not demonstrate improved hypotension outcomes.","source_ids":["S7"]}],"distinctive_claim_remaining":"The remaining differentiator is a patient/session-specific SVD of a multivariable setting-to-physiology response map, followed by outcome-directed modal sensitivity ranking and an out-of-sample reconstruction gate, with prospective success defined jointly as at least 20% fewer symptomatic-hypotension sessions and no more than a 0.1 change in delivered Kt/V. No located source tested that complete algorithm-and-endpoint combination.","confidence":"HIGH"},"implementation_evidence":{"support":"MODERATE","rationale":"Repeated-session patient-specific modeling, real-time prediction, multivariable constrained control, and clinician-reviewed ultrafiltration recommendations have each been implemented. These results make the computational pipeline plausible, but they do not establish stable SVD modes, identifiable independent setting effects, safe counterfactual control selection, or clinical benefit from the combined method.","source_ids":["S5","S6","S7","S8"]},"scores":{"meaningful_impact":{"score":4,"rationale":"Recurrent symptomatic IDH burdens patients and care teams and is associated with important morbidity; a genuine 20% reduction without loss of adequacy would be clinically meaningful.","source_ids":["S1","S2"]},"stakeholder_pull":{"score":2,"rationale":"The general need for individualized IDH prevention is clear, but direct demand for this modal method or a committed adopting site was not found.","source_ids":["S2","S3","S7"]},"incremental_advantage":{"score":2,"rationale":"Multivariable modal control could account for physiology missed by blood-volume-only systems, but no evidence shows that SVD ranking improves outcomes over biofeedback, MPC, prediction alerts, or ordinary individualized prescription review.","source_ids":["S3","S4","S7","S8"]},"distinctiveness_plausibility":{"score":2,"rationale":"The exact SVD/sensitivity/residual-gate combination was not located, but its functional core substantially overlaps established multivariable modeling, optimization, biofeedback, and repeated-session personalization.","source_ids":["S4","S5","S7","S8"]},"technical_implementability":{"score":3,"rationale":"The necessary data streams and control mechanisms exist, but mode stability, identifiability, distribution shift, missing measurements, and safe integration with dialysis machines remain unresolved.","source_ids":["S5","S6","S7","S8"]},"adoption_authority_feasibility":{"score":2,"rationale":"A clinician-in-the-loop advisory pilot is feasible, but operational use affects prescription and potentially device control, requiring nephrologist authority, dialysis-provider governance, machine integration, and likely medical-device regulatory review.","source_ids":["S2","S7"]},"evidence_readiness":{"score":3,"rationale":"Historical session data and measurable endpoints support a retrospective differentiation study and then a randomized crossover trial, although causal evaluation of setting recommendations cannot be completed from observational replay alone.","source_ids":["S3","S5","S6"]},"safety_net_benefit":{"score":4,"rationale":"Clinician review, bounded feasible settings, hard stops, residual rejection, and adequacy constraints could provide useful safeguards if explicitly implemented and validated.","source_ids":["S2","S7"]},"scalability":{"score":3,"rationale":"Software could scale across compatible machines and centers, but heterogeneous data interfaces, local prescriptions, patient populations, and governance would require site-specific validation and integration.","source_ids":["S5","S6","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Retrospective multicenter replay study using existing machine, vital-sign, symptom, intervention, and adequacy data to assess mode stability, reconstruction error, feasible-control identifiability, and differentiation from blood-volume and MPC baselines.","confidence":"LOW","assumptions":["At least two centers can provide harmonized session-level data under existing research agreements.","Work remains offline and does not alter treatment.","The estimate covers data engineering, statistical analysis, clinical review, and governance but not prospective device integration."],"source_ids":["S5","S6","S8"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Single-provider silent-mode implementation followed by clinician-reviewed recommendations, including interfaces, safety bounds, audit logging, cybersecurity review, training, and protocol development.","confidence":"LOW","assumptions":["The system remains advisory and licensed staff authorize every setting change.","Existing machines expose the required data and permit manual implementation.","No new custom sensing hardware is required."],"source_ids":["S7","S8"]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Prospective multicenter randomized crossover evaluation plus production-grade clinical software, quality management, device/vendor integration, monitoring, and regulatory work needed for routine use.","confidence":"LOW","assumptions":["The launch includes several dialysis centers and enough hypotension-prone patients to estimate the 20% outcome target.","Automated actuation, if pursued, materially increases verification and regulatory costs.","The estimate excludes redesign of dialysis-machine hardware."],"source_ids":["S3","S4","S7"]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Model monitoring, recalibration, software maintenance, clinical safety review, incident response, data-interface support, training, and periodic performance audits across an initial provider network.","confidence":"LOW","assumptions":["Deployment is limited to an initial regional or provider network.","Patient-specific modes are periodically re-estimated.","Human review remains part of the operating workflow."],"source_ids":["S5","S6","S7"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"IDH prevalence, clinical burden, and coupled hemodynamic causes are directly supported by systematic-review and official-guidance evidence.","source_ids":["S1","S2"]},"externally_credible_adopter_or_authorizer":{"status":"UNCERTAIN","reason":"Nephrologists and licensed dialysis staff are credible authorizers in the relevant workflow, but no named organization or authority was found willing to evaluate or adopt this specific modal method.","source_ids":["S2","S7"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The hypothesis specifies a differentiated algorithmic package and a measurable comparative endpoint: at least 20% fewer symptomatic-hypotension sessions with delivered Kt/V changing by no more than 0.1.","source_ids":["S3","S4","S5","S8"]},"bounded_next_evidence_step":{"status":"YES","reason":"A bounded retrospective differentiation study can test mode repeatability, reconstruction performance, setting identifiability, and overlap with existing biofeedback/MPC before any treatment recommendation is exposed to clinicians.","source_ids":["S5","S6","S8"]},"no_unresolved_safety_or_authority_stop":{"status":"UNCERTAIN","reason":"The intervention changes prescription-relevant ultrafiltration and dialysate settings within a narrow therapeutic window. Clinician review and hard bounds are plausible controls, but safe limits, override rules, device status, and accountable prescribing authority are not specified in the hypothesis.","source_ids":["S2","S7"]},"credible_cost_scope_and_range":{"status":"UNCERTAIN","reason":"A staged scope can be bounded, but no direct budget evidence was located and cost depends heavily on data access, machine interoperability, advisory-versus-automatic status, and regulatory classification.","source_ids":["S6","S7","S8"]}},"next_evidence_step":"Run a preregistered, multicenter retrospective differentiation study on at least 1,000 sessions from at least 100 patients with recurrent IDH. Fit each patient's response map using prior sessions only; quantify singular-mode stability, spectral separation, out-of-sample reconstruction, and feasible-control sensitivity; and compare proposed recommendations with blood-volume target control and constrained MPC. Proceed to prospective research only if modes are stable, recommendations remain within nephrologist-defined safety and adequacy bounds, and the method adds prespecified predictive or policy value over those close analogues.","blocking_evidence":["No prospective evidence that the SVD/modal policy reduces symptomatic IDH or preserves delivered Kt/V.","No evidence that patient-specific response modes remain identifiable and stable across enough sessions for safe decisions.","No head-to-head differentiation from existing multivariable MPC, blood-volume biofeedback, or clinician-reviewed ultrafiltration controllers.","No externally documented adopter, dialysis-machine partner, or accountable clinical authorizer for the specific method.","Safety bounds, override behavior, sodium and volume balance constraints, and regulatory status are unresolved.","Cost ranges lack direct procurement, integration, trial-budget, or regulatory estimates."],"research_disposition":"PRIOR_ART_DIFFERENTIATION_STUDY","world_novelty_boundary":"This was a targeted open-web literature assessment, not a patent, regulatory-submission, conference-abstract, or exhaustive non-English search. It supports substantial functional collision with multivariable dialysis control and personalized response modeling but does not establish that the exact SVD-plus-sensitivity-plus-reconstruction-gate implementation has or has not appeared anywhere worldwide.","arm":"DISCARDED_AUDIT","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":true,"material_progress_observed":false,"progress_targets":["Demonstrate stable, identifiable patient-specific modes across repeated sessions.","Show incremental value over constrained MPC and established blood-volume biofeedback.","Obtain a named dialysis-provider and nephrologist partner with defined prescribing and override authority.","Specify machine-setting safety constraints, residual rejection thresholds, drift rules, and regulatory pathway.","Prospectively test the preregistered 20% symptomatic-IDH and 0.1 Kt/V endpoint."],"reason":"As written, the hypothesis does not satisfy the strict differentiated-opportunity endpoint. The problem and technical premise are credible, and the residual claim is testable, but the functional core substantially collides with established multivariable and biofeedback dialysis control. Incremental clinical advantage, mode stability, adopter commitment, safety authority, and credible cost evidence remain unverified. The early discard therefore is not shown to be a false negative on current external evidence."}}