{"id":"73338886-98a0-42a5-914c-4f02eab89862","arxiv_id":"2605.12891","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":2,"one_line_summary":"DMAT is a new intent-driven optimization framework for VMAT that jointly optimizes dosimetry, delivery time, and complexity using machine emulation and adaptive control point allocation on H&N, lung, and prostate cases.","lead":"The paper introduces Dynamic Modulated Arc Therapy (DMAT), a framework that co-optimizes radiation dose quality, delivery time, and modulation complexity in arc therapy by incorporating machine dynamics and user-selected modulation levels. This could allow clinicians to make informed trade-offs in treatment planning for time-sensitive radiotherapy cases.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Delivery time accuracy claim rests on unvalidated hypothetical machine emulation without real-linac comparison","rationale":"The reader's weakest_assumption matches the load-bearing point exactly. Full-text evaluation confirms all timing results derive from the same unvalidated hypothetical model, so the central claim cannot be accepted without the proposed real-machine check. This does not invalidate the algorithmic framework but conditions acceptance on external validation.","tokens_in":1848,"tokens_out":336,"duration_ms":16812,"concrete_test":"For the three H&N plans at modulation level +2, export the final control-point sequences and MU values; deliver them on a clinical linac whose maximum speeds are within 10% of the hypothetical values; compare measured beam-on plus gantry/MLC transit times against the DMAT-predicted times; if mean absolute deviation exceeds 12% the accuracy assertion is unsupported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The strongest claim requires that DMAT exposes 'accurate delivery time' via machine-aware timing so that quality-complexity trade-offs become transparent. All reported results (H&N, lung SBRT, prostate SBRT) are generated exclusively with a fixed hypothetical linac (2.5 RPM gantry, 6.25 cm/s MLC, 3000 MU/min) and an internal emulation of axis synchronization and acceleration. No section describes calibration of the emulation against measured control-point logs, delivery-time measurements on a physical accelerator, or sensitivity of predicted times to parameter perturbations. Consequently the 'accurate' qualifier and the navigability claim for motion-sensitive or adaptive workflows rest on an untested modeling assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper introduces Dynamic Modulated Arc Therapy (DMAT), an intent-driven framework for VMAT planning that co-optimizes dosimetric quality, delivery time, and modulation complexity. It couples a machine-emulation model (accounting for axis synchronization and finite acceleration) with dynamic modulation control via a user-selected level (-3 to +3) that governs leaf travel, MU behavior, and CP density. Plans are generated by initializing CP geometry then iteratively alternating dosimetric updates with sequencing penalties, followed by adaptive CP redistribution. Evaluation on H&N, lung SBRT, and prostate SBRT cases with a hypothetical linac (2.5 RPM gantry, 6.25 cm/s MLC, 3000 MU/min) shows higher modulation levels increase MU/Gy and delivery time while adaptive allocation concentrates resolution in complex sectors; negative levels shorten time at some quality cost. The central claim is that exposing accurate delivery time during planning makes quality-complexity trade-offs transparent for time-constrained workflows.","tokens_in":1986,"tokens_out":647,"duration_ms":18239,"significance":"If the machine-emulation model were shown to predict real delivery dynamics, DMAT would address a recognized gap in traditional VMAT by treating delivery time as an explicit, steerable parameter rather than an emergent property. This could support planning for motion-sensitive or adaptive radiotherapy where timing trade-offs must be navigated explicitly. The adaptive CP allocation and modulation-level control are novel elements that could be useful if quantitatively validated.","major_comments":[{"comment":"Abstract and evaluation section: The claim that DMAT 'exposes accurate delivery time' and makes trade-offs 'transparent and navigable' for motion-sensitive workflows rests entirely on an unvalidated hypothetical machine model (2.5 RPM gantry, 6.25 cm/s MLC, 3000 MU/min) with internal emulation of synchronization and acceleration. No section reports calibration against measured control-point logs, physical delivery-time measurements, or sensitivity of predicted times to parameter perturbations.","section":"Abstract; evaluation on H&N, lung SBRT, prostate SBRT cases"},{"comment":"Evaluation section: Reported results are limited to qualitative statements (e.g., 'substantial quality gains' for H&N, 'smaller incremental improvements' for SBRT cases) without quantitative dosimetric metrics (DVH values, conformity indices), delivery-time numbers, MU/Gy values, or direct comparisons to standard VMAT plans, preventing assessment of effect sizes or clinical relevance.","section":"Evaluation on H&N, lung SBRT, and prostate SBRT cases"}],"minor_comments":[{"comment":"The modulation level parameter (-3 to +3) is introduced without an explicit mathematical definition or mapping to the penalties for motion, velocity, MU uniformity, and complexity.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The manuscript presents a conceptual framework without any empirical validation data against real linac delivery; this raises questions about whether the work meets the typical threshold for a methods paper in medical physics without at least a small validation cohort."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive and detailed comments. We agree that the manuscript requires revisions to clarify the hypothetical nature of the machine model and to strengthen the evaluation with quantitative data. Our point-by-point responses follow.","responses":[{"response":"We acknowledge that the machine-emulation model is hypothetical and uncalibrated against real control-point logs or measured delivery times. The phrasing 'exposes accurate delivery time' in the abstract and conclusion overstates the current evidence. In the revised manuscript we will replace 'accurate' with 'predicted' throughout, explicitly note the lack of empirical calibration, and add a limitations paragraph discussing sensitivity to parameter perturbations. These changes will temper the claims regarding immediate applicability to motion-sensitive workflows while preserving the framework's conceptual contribution.","revision_made":"yes","referee_comment":"[Abstract; evaluation on H&N, lung SBRT, prostate SBRT cases] Abstract and evaluation section: The claim that DMAT 'exposes accurate delivery time' and makes trade-offs 'transparent and navigable' for motion-sensitive workflows rests entirely on an unvalidated hypothetical machine model (2.5 RPM gantry, 6.25 cm/s MLC, 3000 MU/min) with internal emulation of synchronization and acceleration. No section reports calibration against measured control-point logs, physical delivery-time measurements, or sensitivity of predicted times to parameter perturbations."},{"response":"The current evaluation section presents results qualitatively. We agree that quantitative metrics are required to allow assessment of effect sizes. In the revised manuscript we will expand the evaluation to report specific DVH endpoints, conformity indices, delivery times, MU/Gy values, and side-by-side comparisons against standard VMAT plans for all three anatomical sites. These additions will be placed in a new results subsection with accompanying tables.","revision_made":"yes","referee_comment":"[Evaluation on H&N, lung SBRT, and prostate SBRT cases] Evaluation section: Reported results are limited to qualitative statements (e.g., 'substantial quality gains' for H&N, 'smaller incremental improvements' for SBRT cases) without quantitative dosimetric metrics (DVH values, conformity indices), delivery-time numbers, MU/Gy values, or direct comparisons to standard VMAT plans, preventing assessment of effect sizes or clinical relevance."}],"tokens_in":1662,"tokens_out":496,"duration_ms":29655,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The new element here is the modulation level knob from -3 to +3 that directly steers leaf travel, MU, and CP density while the optimizer alternates dosimetric updates with sequencing penalties and then redistributes control points toward high-complexity sectors. That combination, plus the explicit machine emulation for gantry and MLC dynamics, is not standard in the VMAT literature the abstract cites.\n\nThe framework description is clear on how intent-driven planning is supposed to work: higher levels increase MU/Gy and time on H&N cases, lower levels cut time on prostate and lung SBRT at some quality cost. The adaptive CP step is a practical addition that avoids wasting resolution on easy arc sectors.\n\nThe soft spot is exactly where the stress-test note flags it. All timing numbers come from one fixed hypothetical linac (2.5 RPM, 6.25 cm/s MLC, 3000 MU/min) with no calibration against measured logs, no delivery-time measurements on a real accelerator, and no sensitivity checks. Without that, the claim that the method exposes “accurate” delivery time for motion-sensitive or adaptive workflows stays unproven. The reported results are also only qualitative; there are no dose-volume metrics, no plan-quality scores, and no head-to-head numbers against conventional VMAT.\n\nThis is for optimization researchers who want to experiment with time as an explicit objective. A reader already working on VMAT sequencing might pick up the modulation-level idea and the CP-redistribution trick. Anyone needing evidence that the times are reliable for clinical decisions would have to wait for more data.\n\nIt is worth sending for peer review once the authors add at least one real-linac comparison or a sensitivity study on the machine parameters. The core machinery is distinct enough that referees could give useful feedback on whether the approach scales.","headline":"DMAT adds user-selectable modulation levels and adaptive CP allocation to VMAT planning but its timing claims rest on an untested hypothetical machine model.","tokens_in":2530,"tokens_out":443,"would_cite":false,"duration_ms":21662,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"DMAT co-optimizes radiotherapy plan quality, delivery time, and modulation complexity by exposing machine-aware timing during planning.","keywords":["DMAT","VMAT optimization","radiotherapy planning","delivery time","modulation complexity","arc therapy","machine emulation","adaptive radiotherapy"],"falsifier":"A direct comparison of DMAT-predicted delivery times against measured times on a clinical linear accelerator operating at the assumed speeds would falsify the claim if discrepancies exceed acceptable clinical margins.","tokens_in":2765,"feed_emoji":"⏱️","tokens_out":434,"duration_ms":20826,"temperature":0.7,"pith_summary":"The paper introduces Dynamic Modulated Arc Therapy (DMAT) as a framework that treats delivery time as a steerable parameter rather than an emergent property in arc therapy planning. It couples machine emulation of dynamic limits with optimization that allows users to select modulation levels controlling leaf travel, MU, and control point density. This makes trade-offs between plan quality and treatment duration explicit and navigable. A sympathetic reader would care because it supports time-constrained workflows such as motion-sensitive or adaptive radiotherapy on next-generation systems.","feed_headline":"DMAT turns delivery time into a planning choice in radiotherapy","feed_subtitle":"Co-optimizing quality and complexity with machine timing makes trade-offs transparent for motion-sensitive treatments.","key_machinery":"Intent-driven modulation level (-3 to +3) coupled with machine emulation of axis synchronization and finite acceleration, plus iterative dosimetric and sequencing updates.","core_discovery":"DMAT jointly co-optimizes dosimetric quality, delivery time, and modulation complexity through machine emulation accounting for axis synchronization and finite acceleration, dynamic modulation control, and clinical cost functions, with a user-selected level from -3 to +3 governing behavior, and adaptive control point allocation, as demonstrated on H&N, lung SBRT, and prostate SBRT cases using a hypothetical machine.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["DMAT makes delivery time a steerable planning parameter","DMAT co-optimizes quality with delivery time and modulation","Machine emulation in DMAT balances dose and arc duration","User levels in DMAT govern leaf travel MU and CP density"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The hypothetical machine parameters and the machine emulation model accurately represent real-world delivery dynamics and synchronization.","fun_headline_variants_meta":{"raw":{"variants":["DMAT makes delivery time a steerable planning parameter","DMAT co-optimizes quality with delivery time and modulation","Machine emulation in DMAT balances dose and arc duration","User levels in DMAT govern leaf travel MU and CP density"]},"model":"grok-4.3","cost_usd":0.007957,"raw_usage":{"total_tokens":3679,"prompt_tokens":777,"num_sources_used":0,"completion_tokens":51,"cost_in_usd_ticks":79574500,"prompt_tokens_details":{"text_tokens":777,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2851,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":777,"tokens_out":51,"duration_ms":21450,"temperature":1.0,"reasoning_tokens":2851,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-30T21:30:42.086201+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A direct comparison of DMAT-predicted delivery times against measured times on a clinical linear accelerator operating at the assumed speeds would falsify the claim if discrepancies exceed acceptable clinical margins.","supporting_citations":[],"review_version":1}