Writing study: PhenoBrain: Phenotype-Conditioned Long-Range Communication for Multi-Modal Brain Network Analysis

ICML overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICML 2026 oral writing corpus

Lingyuan Meng, Ke Liang, Hao Li, Meng Liu, Weijia Shi, Miaomiao Li, Yang Gao, Xinwang Liu · Official ICML 2026 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Domain-grounded methods & physical systems (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

13 physical pages

Before references / appendix

≈ 8.8 page-equivalents

Main figures / tables

6 / 1

Attached appendix figures / tables

2 / 0

Attached appendix span

≈ 2.91 page-equivalents

Before references / appendix: ≈ 67.69%References / intervening material: ≈ 9.92%Attached appendix: ≈ 22.38%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	1	≈ 2.47 pp.	19%
2. Methodology	3	≈ 2.82 pp.	21.69%
3. Experiments	6	≈ 2.71 pp.	20.85%
4. Conclusion	9	≈ 0.34 pp.	2.62%
Impact Statement	9	≈ 0.11 pp.	0.85%
References	9	≈ 1.28 pp.	9.85%
A. Appendix	11	≈ 2.91 pp.	22.38%

All detected figure and table locations

Main text

• Figure 1 · p. 2
• Figure 2 · p. 4
• Table 1 · p. 7
• Figure 3 · p. 7
• Figure 4 · p. 8
• Figure 5 · p. 8
• Figure 6 · p. 9

Attached appendix

• Figure 7 · p. 12
• Figure 8 · p. 12

Argument and organization

1. 1 Introduction and Related Work

   Distinguishes phenotype-guided computation from late feature fusion and motivates long-range graph communication.

2. 2 Methodology

   Defines input preparation, phenotype context, routing, prompts, modulated attention, final fusion and losses in separate subsections.

3. 3 Experiments; 4 Conclusion

   Presents task metrics, hop-length response, component removals and routing/attention visualizations.

Problem, claim and contribution

The central distinction is where phenotype enters the computation: it conditions routing and attention before the final fusion head.

Sections 1–2.2, physical pp.1–4 · Main paper

Argument and section order

The method orders data construction, modality encoding, routing/attention and training, allowing the information available at each stage to be traced.

Sections 2.3–2.6, pp.4–6 · Main paper

Figures, captions and visual explanation

The framework draws separate phenotype paths into routing, prompts and attention; the result page pairs the comparison table with hop-length curves and preserves an ABIDE sensitivity non-win.

Figure 2, p.4; Table 1 and Figure 3, p.7 · Main paper

Tables, comparisons and evidence

The source differentiates ASD classification on ABIDE from child/adult classification on PIXAR in the appendix, even though the main discussion uses broader disease-prediction language.

Appendix A.2.1, p.11; Section 3.2, p.7 · Main paper

Appendix and reproducibility

The short appendix repeats interpretation figures and provides broad preprocessing/hardware descriptions, but does not fulfill the main text’s promise of complete split, model and hyperparameter details.

Section 3.1, p.6; Appendix A.2, pp.11–13 · Main paper

Clarity and writing conventions

Attention-based interpretation is reported at network, ROI and bootstrap-overlap levels; this operationalizes stability but is not equivalent to identifying a causal biomarker.

Figure 6 and Section 3.4 continuation, p.9; Appendix A.2.4, pp.12–13 · Main paper

What to reuse—and what not to infer

Transferable lesson: Draw every conditioning path explicitly and distinguish predictive evidence, model-derived explanations and their stability; check that appendix promises are actually fulfilled.

Boundary: The supplied source does not establish clinical validity or causal neural communication. PIXAR is a developmental classification task, not a second disease cohort. Text-diagnosis keyword removal is claimed in Section2.3, but is not a comprehensive leakage audit; the Figure2 illustration still includes an ASD diagnosis phrase. Cohort/split and detailed hyperparameter reporting gaps limit reproducibility.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-10-07

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Writing-focused scan across physical PDF pp.1–9, covering framing, section progression, definitions, method/evidence explanations, captions and scope. Physical pages extracted individually to avoid form-feed pagination errors. Selected full-page rasters inspected; not exhaustive line-by-line technical verification or experimental reproduction.

Appendix / supplement

All appendix pages11–13 read for writing/reproducibility scope, including A.1 related work, A.2 experimental details and routing/attention interpretation. Dataset partitioning and many numeric implementation settings promised by the main text are not supplied there. No clinical or biological validation was performed.

Visually inspected PDF pages

4, 7

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Published proceedings PDF, PMLR volume 306, retrieved 2026-10-07 · 13 pages