{"id":"99153800-73f8-41a7-9d4b-214168b86839","arxiv_id":"2607.11656","paper_version":2,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":5.5,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"An imputation-free transformer with masked and intersample attention predicts Alzheimer’s status and scores across cohorts with better calibration than tree ensembles.","lead":"NITROGEN is an imputation-free transformer that predicts Alzheimer’s diagnosis and severity from incomplete clinical records while reporting calibrated uncertainty. It matters because real-world AD data are messy and missing, and overconfident models are unsafe for clinical use.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"Abstract-only review cannot verify the load-bearing claim that masked+intersample attention yields genuine calibration gains under cohort shift rather than artifacts of missingness patterns or evaluation design.","rationale":"The Reader’s weakest_assumption correctly isolates the load-bearing condition: that masked + intersample attention can extract transferable clinical structure without a formal missingness model, and that calibration gains survive cohort shift for non-artifactual reasons. With only the abstract available, no stronger or different concern can be substantiated; the evaluation design (ADNI train, OASIS-3/AIBL external) is methodologically sensible, and the problem is real, so the CONDITIONAL / LOW-confidence verdict is appropriate. No formal verification or code is present to raise or lower the bar. The concrete test above would settle whether the concern lands once full artifacts exist. Verdict remains CONDITIONAL pending those checks; no adjustment is warranted from an abstract-only second pass.","tokens_in":2094,"tokens_out":546,"duration_ms":4654,"concrete_test":"Obtain the full paper (or code/supplement) and recompute ECE / reliability diagrams and AUROC for NITROGEN vs. the same tree ensembles on OASIS-3 and AIBL under (i) the paper’s native missingness encoding and (ii) a controlled MAR simulation that matches ADNI missing rates; if calibration advantage disappears or reverses under matched missingness, the gains are likely artifacts of cohort missingness patterns rather than the attention mechanism.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that NITROGEN’s masked within-patient attention plus intersample attention learns clinically valid structure from partially observed multimodal records without a formal missingness model (MAR/MNAR), and that the reported calibration/uncertainty advantages over tree ensembles on OASIS-3 and AIBL are not artifacts of cohort-specific missingness, modality-marking conventions, or how “absent” is encoded. The abstract asserts robust calibration and competitive discrimination under shift, plus a modality-aware uncertainty adjustment, but supplies no quantitative calibration metrics (ECE, Brier, reliability diagrams), no missingness-mechanism analysis, no ablation of intersample attention, and no description of how modalities are marked absent or how baselines handle missingness. Without those, the transfer of structure from ADNI to external cohorts—and therefore the claim that imputation-free attention is the source of the gains—remains uncheckable. This is the same soft spot the Reader flagged; it is load-bearing because the paper’s contribution is precisely the claim that this architecture avoids imputation bias while remaining calibrated under real-world incompleteness.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript proposes NITROGEN, an imputation-free transformer that models within-patient feature dependencies and between-patient relational structure via masked and intersample attention, for Alzheimer’s diagnostic classification and cognitive-score prediction from partially observed multimodal records. It is trained on ADNI (N=7858 scans) and evaluated on external cohorts OASIS-3 (N=2675) and AIBL (N=1286). The abstract claims robust calibration and uncertainty-quantification advantages over tree-based ensembles under cohort shift, competitive discrimination, post-hoc identification of temporal-pole thickness, age, and APOE as important features, and a modality-aware uncertainty adjustment that scales predictive uncertainty with the importance of absent modalities.","tokens_in":2408,"tokens_out":882,"duration_ms":16870,"significance":"If substantiated with quantitative evidence, the work would address a clinically material problem: imputation-induced bias and overconfidence under real-world incompleteness in AD modelling. Emphasizing calibration, cross-cohort reliability, and uncertainty—not accuracy alone—is appropriate for deployment-oriented clinical ML. The multi-cohort external-test design and the explicit modality-aware uncertainty mechanism are, in principle, strengths. Significance hinges on whether the reported calibration gains under shift are real, transferable, and attributable to the proposed attention design rather than evaluation or missingness artifacts.","major_comments":[{"comment":"The central claim of robust calibration and uncertainty advantages under cohort shift cannot be assessed from the abstract alone: no ECE, Brier score, reliability diagrams, confidence intervals, or statistical comparisons versus tree ensembles are reported. These metrics are load-bearing for the paper’s contribution and must appear with full experimental detail in the manuscript.","section":null},{"comment":"The architecture claim rests on masked within-patient plus intersample attention learning clinically valid structure without a formal missingness model (MAR/MNAR). The abstract does not state how absent modalities are marked, what missingness rates or patterns exist across ADNI/OASIS-3/AIBL, or whether gains could be artifacts of cohort-specific incompleteness. A missingness analysis and encoding description are required.","section":null},{"comment":"No ablation of intersample (between-patient) attention versus masked-only attention is described. Because intersample attention is an invented, load-bearing component and carries leakage/cohort-artifact risk, its incremental contribution to discrimination and calibration on external cohorts must be quantified.","section":null},{"comment":"Baseline fairness is unclear: tree ensembles often handle missingness natively, whereas the abstract contrasts NITROGEN with imputation-based practice. The manuscript must specify exactly how each baseline treats missing values and ensure the comparison isolates the proposed imputation-free design rather than differing missingness handling.","section":null}],"minor_comments":[{"comment":"Abstract phrasing ‘N=7858 scans’ (and analogous for external cohorts) should be clarified as subjects vs. visits/scans to avoid double-counting concerns in longitudinal AD cohorts.","section":null},{"comment":"Feature-importance statements (temporal pole thickness, age, APOE) should be tied to a named method (e.g., attention rollout, permutation, SHAP) and to task (classification vs. score prediction) once full results are presented.","section":null},{"comment":"The modality-aware uncertainty adjustment is introduced only at a high level; coefficients, how modality importance is estimated, and whether adjustment is post-hoc or trained should be specified for reproducibility.","section":null}],"recommendation":"uncertain","confidential_remarks":"This is an abstract-only review; full text, tables, and figures were not available. I cannot verify load-bearing quantitative claims (calibration metrics, ablations, missingness handling). Recommendation is therefore uncertain pending the complete manuscript. If the full paper supplies ECE/Brier/reliability results, missingness analysis, intersample ablations, and fair baseline missingness protocols, the contribution could be reassessed toward minor or major revision rather than rejection. Scope (q-bio.NC / clinical ML for AD) appears appropriate for a methods-oriented journal in the field."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"Punchline: this is a cleanly framed clinical-ML paper on a real deployment barrier—missing multimodal AD data and overconfident models—but we only have the abstract, so the load-bearing claim (imputation-free attention yields genuine calibration under cohort shift) is still uncheckable.\n\nWhat is actually new is the package, not any single piece. Masked attention for missing features and intersample attention both have prior art; the contribution is applying them jointly as NITROGEN on ADNI (N=7858) with external tests on OASIS-3 and AIBL, plus a modality-aware uncertainty adjustment that scales confidence by the importance of absent modalities. The evaluation story is the right one: train on ADNI, hold out two independent cohorts, and report calibration and uncertainty alongside discrimination rather than accuracy alone. Feature importance (temporal pole thickness, age, APOE) is presented as post-hoc and not individually sufficient, which is honest.\n\nSoft spots are proportional to what we can see. Without ECE/Brier numbers, reliability diagrams, missingness rates, ablations of intersample attention, or a clear account of how “absent” is encoded and how tree ensembles handle missingness, we cannot tell whether the reported calibration edge is real or an artifact of cohort-specific missingness patterns. The abstract asserts robust calibration under shift and expected degradation on more distant cohorts; that is plausible, but free parameters (architecture, uncertainty scaling) and the lack of a formal MAR/MNAR discussion leave the transfer story open. Circularity burden looks low for this empirical setup; the risk is unreproducible detail and baseline under-comparison, not tautology.\n\nWho it is for: people building multimodal AD models or tabular transformers under real-world incompleteness. It deserves a serious referee if the full paper ships the metrics, ablations, and missingness analysis the abstract promises. I would not cite or bring it to reading group on the abstract alone. Send it to peer review rather than desk-reject; the problem and design justify referee time even if revision will be heavy on quantification.","headline":"Sensible multi-cohort AD setup for a real missing-data problem, but abstract-only so the calibration gains remain unverified claims.","tokens_in":3022,"tokens_out":515,"would_cite":false,"duration_ms":9030,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"An imputation-free transformer predicts Alzheimer’s diagnosis and cognitive scores with calibrated uncertainty from incomplete clinical records across cohorts.","keywords":["Alzheimer's disease","imputation-free learning","transformer","uncertainty quantification","multimodal clinical data","cohort shift","calibration","masked attention"],"falsifier":"On a multi-site held-out cohort with systematically different missingness (for example, imaging or APOE routinely absent), if NITROGEN’s calibration (ECE or reliability diagrams) is no better than a carefully imputed tree ensemble while discrimination also collapses, the central claim fails.","tokens_in":2990,"feed_emoji":"🧠","tokens_out":807,"duration_ms":15900,"temperature":0.7,"pith_summary":"Incomplete, heterogeneous clinical data block reliable Alzheimer’s diagnosis and severity prediction, and conventional imputation adds bias and overconfidence. This paper introduces NITROGEN, a transformer that never fills missing values; instead it uses masked attention inside each patient and intersample attention across patients so it can learn directly from partially observed multimodal records. Trained on ADNI and tested on independent OASIS-3 and AIBL cohorts, NITROGEN remains competitively discriminative while showing clearer calibration and uncertainty advantages over tree-based ensembles under cohort shift. A modality-aware uncertainty adjustment further raises predictive uncertainty when important modalities are absent. If the claim holds, clinical AI for Alzheimer’s should be judged on calibration, interpretability, and cross-cohort reliability—not accuracy alone.","feed_headline":"Imputation-free transformer predicts Alzheimer's with calibrated uncertainty","feed_subtitle":"Trained on ADNI, it keeps calibration and discrimination on OASIS-3 and AIBL without filling missing values.","key_machinery":"Masked and intersample attention inside the NITROGEN transformer: masked attention lets the model use whatever features are observed for a given patient, while intersample attention captures relational structure across patients, so missing values never need to be imputed.","core_discovery":"NITROGEN, an imputation-free transformer that jointly models within-patient feature dependencies via masked attention and between-patient structure via intersample attention, enables robust Alzheimer’s diagnostic classification and cognitive-score prediction with calibrated uncertainty across heterogeneous cohorts, without the systematic bias of conventional imputation.","pith_inferences":["If attention recovers valid structure without a missingness model, similar architectures could apply to other chronic diseases with sparse multi-site health records.","Importance-weighted missingness penalties like the modality-aware adjustment could improve trustworthiness of any multimodal medical predictor.","External validation on community versus research clinics with deliberately mismatched missingness would test whether calibration gains are method-driven or data-pattern-driven.","Pairing the learned feature importances with longitudinal trajectories could yield earlier risk scores that stay calibrated as more modalities appear over time."],"forward_implications":["Clinical models can train and deploy on incomplete multimodal records without an imputation step that distorts feature relationships.","Predictive confidence can be raised when high-importance modalities are absent, giving clinicians a usable uncertainty signal.","Performance drop on more distributionally different cohorts becomes a measurable reliability check rather than a silent failure.","Cortical thickness in the temporal pole, age, and APOE genotype are consistently informative for AD classification, though none is sufficient alone.","Diagnostic AI evaluation protocols should report calibration and cross-cohort transfer alongside accuracy."],"fun_headline_variants":["NITROGEN predicts Alzheimer's without imputation, stays calibrated across cohorts","Imputation-free transformer models AD from incomplete data via masked attention","Transformer learns AD diagnosis and scores without filling missing clinical values","NITROGEN holds uncertainty calibration for Alzheimer's on OASIS-3 and AIBL","Within- and between-patient attention enables robust AD prediction without imputation"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"Masked within-patient attention plus intersample attention can learn clinically valid structure from partial records without any formal model of why data are missing, and that structure transfers across cohorts whose missingness patterns differ.","fun_headline_variants_meta":{"raw":{"variants":["NITROGEN predicts Alzheimer's without imputation, stays calibrated across cohorts","Imputation-free transformer models AD from incomplete data via masked attention","Transformer learns AD diagnosis and scores without filling missing clinical values","NITROGEN holds uncertainty calibration for Alzheimer's on OASIS-3 and AIBL","Within- and between-patient attention enables robust AD prediction without imputation"]},"model":"grok-4.5","effort":"low","cost_usd":0.0053,"raw_usage":{"total_tokens":1498,"prompt_tokens":826,"num_sources_used":0,"completion_tokens":76,"cost_in_usd_ticks":53000000,"prompt_tokens_details":{"text_tokens":826,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":596,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":826,"tokens_out":76,"duration_ms":4499,"temperature":1.0,"reasoning_tokens":596,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-15T08:43:01.933406+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"On a multi-site held-out cohort with systematically different missingness (for example, imaging or APOE routinely absent), if NITROGEN’s calibration (ECE or reliability diagrams) is no better than a carefully imputed tree ensemble while discrimination also collapses, the central claim fails.","supporting_citations":[],"review_version":2}