Pith. sign in

REVIEW 4 major objections 5 minor 136 references

The paper argues that a supervisory AI agent coordinating seven biological-level agents can model human physiology across scales, enabling long-term drug efficacy and toxicity prediction beyond localized models.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

The paper proposes, but does not implement or validate, a multi-agent AI framework for cross-scale modeling of human biology from molecules to whole body, with sketches of metastasis scoring and drug development.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A serious, well-organized roadmap for multi-scale biomedical AI agents, but the predictive claims are not backed by implementation; the temporal-alignment problem is admitted and load-bearing. the 4 major comments →

arxiv 2508.19800 v1 pith:K4KIHDVD submitted 2025-08-27 q-bio.TO q-bio.BM

A Multi-Layered Framework for Modeling Human Biology: From Basic AI Agents to a Full-Body AI Agent

Aoqi Wang , Jiajia Liu , Jianguo Wen , Yangyang Luo , Zhiwei Fan , Liren Yang , Xi Hu , Ruihan Luo
show 4 more authors
Yankai Yu Sophia Li Weiling Zhao Xiaobo Zhou
This is my paper
classification q-bio.TO q-bio.BM
keywords Full-Body AI Agentmulti-agent systemscross-scale reasoningsystems biologytumor metastasis scoringdrug developmentorganoids and organ-on-chiplarge language models
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proposes a Full-Body AI Agent: a supervisor AI that coordinates seven specialized agents—molecule, organelle, cell, tissue, organ, organ system, and body system—to model human biology from molecular to whole-body scales. The central claim is that because the agents exchange results iteratively and bidirectionally, the system can predict long-term drug efficacy and toxicity better than models confined to one biological level. The framework is demonstrated in two cases: a metastasis AI Agent that scores tumor progression in three phases, and a drug AI Agent that guides organoid and chip-based preclinical models under whole-body physiological constraints. A worked example computes an Initiation Score from snRNA-seq data for a lung cancer sample, showing one phase in practice. A sympathetic reader would care because the proposal directly targets the translational gap where molecular findings fail to predict systemic outcomes.

Core claim

The central discovery is architectural: a multi-level biological problem can be decomposed into tasks assigned to seven biology-grounded agents, whose outputs are then re-integrated through iterative bidirectional exchange. The Full-Body AI Agent acts as both supervisor and integrator, perceiving multi-modal data, generating cross-scale hypotheses, decomposing problems, and looping conclusions from one biological level back into others until a physiologically plausible whole-body model converges. On this basis the paper claims predictive power for long-term efficacy and toxicity that localized models lack, and it instantiates the claim in a three-phase metastasis scoring system and a drug-de

What carries the argument

The load-bearing mechanism is the hierarchical multi-agent loop: a supervisory Full-Body AI Agent plus seven basic agents, each bound to one biological scale, communicating through a standardized Data Commons. The loop enforces bidirectional constraints—molecular and cellular findings propagate upward to tissue, organ, and system levels, while systemic plausibility filters propagate downward to refine or reject lower-level explanations. The implemented demonstrations are the Metastasis AI Agent's initiation/dissemination/colonization scores and the Drug AI Agent's full-body physiological wrapping of preclinical organoid and chip models.

Load-bearing premise

The framework assumes that heterogeneous biological data at different scales can be standardized and temporally aligned into a single reasoning loop; the paper itself flags temporal alignment as largely unsolved.

What would settle it

Take a retrospective cohort with long-term follow-up and compute the three-phase metastasis scores from the same multi-scale data; if the scores do not predict metastatic progression beyond standard staging or beyond a molecular-only score, the cross-scale advantage fails. Likewise, a prospective evaluation of the Drug AI Agent's predicted organ-level toxicities against clinical trial outcomes would settle the efficacy/toxicity claim.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the framework works, cancer metastasis risk could be scored phase-by-phase (initiation, dissemination, colonization), letting clinicians target stage-specific vulnerabilities.
  • Drug developers could place organoid and organ-on-chip results inside whole-body constraints, catching long-term and distal toxicities before clinical trials.
  • Molecular discoveries would no longer be interpreted in isolation; every finding could be traced through tissue, organ, and system effects.
  • The same architecture could reduce the more-than-90% attrition of drug candidates by exposing system-level failures earlier in the pipeline.
  • Standardized data commons would make multi-omics, imaging, and clinical data interoperable across the seven biological levels.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The three-phase metastasis scores look like a testable prognostic instrument: if they beat conventional staging in retrospective cohorts, they could become a clinical biomarker; the paper does not itself run that validation.
  • If temporal alignment is solved, the same supervisor architecture could evolve into a continuous digital twin of an individual patient, updating predictions as new clinical and wearable data arrive.
  • The 'whole-body constraints' idea suggests a new reporting standard for in vitro models: every organoid or chip result should include a statement of which systemic constraints it may violate.
  • The seven-agent decomposition could be transferred to non-human species or even multi-organism systems by swapping level-specific agents, a direction the paper leaves implicit.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This paper proposes a multi-agent AI architecture, the "Full-Body AI Agent," consisting of a supervisory agent and seven biological-level agents (molecule, organelle, cell, tissue, organ, organ system, body system). The authors describe a reasoning pipeline spanning data perception, hypothesis generation, task decomposition, execution, and iterative multi-scale feedback, and they propose two applications: a Metastasis AI Agent with Initiation, Dissemination, and Colonization scores, and a Drug AI Agent intended to couple molecular discovery with organoid/organ-on-chip models under whole-body constraints. The only quantitative exercise is an Initiation Score computed from a single NSCLC snRNA-seq sample; the Tissue and Organ agents are not applied. The abstract claims that the approach "enables the predictive modeling of long-term efficacy and toxicity beyond what localized models alone can achieve."

Significance. If the framework were implemented and validated, it could meaningfully advance multi-scale biomedical modeling by formalizing cross-level integration and providing an organizing substrate for heterogeneous data. The paper's strengths are its systematic decomposition of biological levels, its broad literature synthesis, and the reasonable three-phase metastasis scoring concept. However, the manuscript provides no implementation, no end-to-end validation, and no comparison against localized models. The one worked example is a static, single-sample analysis with a partly circular design, and the paper itself concedes in §7 that temporal alignment of heterogeneous datasets is "still a largely unsolved problem." Thus the central predictive claim is not supported by the evidence presented. As a vision or perspective, the paper has heuristic value, but as a research claim it falls short of the stated scope.

major comments (4)
  1. [§5, Case 1 (Initiation Score)] The only quantitative demonstration does not support the claim to predict metastatic potential beyond localized models. It analyzes one static snRNA-seq sample (N2254) and computes an Initiation Score using curated EMT, stemness, and metabolic gene sets with hand-set thresholds (top 20% hybrid EMT; top 25% high-initiation, threshold 0.324) and PCA first-component weights. Cells classified as high-initiation are then used in differential expression analysis that reports upregulation of CD44, FN1, and VIM; these are expected components of the very EMT/stemness/invasion programs used to construct the score, so the "validation" is partly circular. The authors also state that the Tissue AI Agent and Organ AI Agent were not applied, so the exercise does not demonstrate cross-scale integration. No clinical endpoint, independent cohort, or comparison with existing localized predictors is provide
  2. [§7 and §3.4] The central abstract claim concerns "long-term efficacy and toxicity," which is inherently longitudinal. Yet §7 explicitly states that "the temporal alignment of heterogeneous datasets, particularly for dynamic physiological processes, is still a largely unsolved problem." The bidirectional feedback loop in §3.4 requires outputs from one biological level to be iteratively exchanged with others, but no mechanism, formalism, or data structure for temporal alignment is defined or demonstrated. The only worked example is a static snapshot. Without a temporal reference frame, the system cannot propagate information through time, and the claimed advantage over localized models is therefore unsupported. This is a load-bearing gap, not a peripheral caveat.
  3. [§5, Case 2 (Drug AI Agent)] The drug development case study is a tool enumeration rather than a demonstration. The text lists AutoDock-GPU, GROMACS, pkCSM, hERG-Block, Tox21, retrosynthesis tools, and organoid/organ-on-chip concepts, but no end-to-end agent run, no quantitative prediction of efficacy or toxicity, and no benchmark against standard preclinical models is reported. The cited successes (ISM001-055, Halicin, Baricitinib) are external discoveries made by other systems and cannot serve as evidence for the proposed framework. The conclusion that the Drug AI Agent "can transcend conventional siloed stages" is therefore a hope, not a result.
  4. [§3.5 and §6.1] The claimed comparative advantage over existing multi-agent systems rests on the reliability of LLM-based agents to perceive, reason, and execute domain-specific biological analyses, and on convergence of the iterative cross-scale loop without amplifying errors. No evidence is provided for either. Section 6.1 itself lists data integration, interpretability, scalability, and data quality as open challenges, but the paper does not show how the framework mitigates them. Until at least a pilot implementation with end-to-end results is presented, the assertion that this architecture outperforms localized models or existing multi-agent systems remains a conjecture.
minor comments (5)
  1. [Figure 10 legend] The legend reads "Predefined detachable biological tasks for the Organ System AI Agent," but Section 4.7 describes the Body System AI Agent. Please correct the mismatch.
  2. [References [53] and [54]] Reference [53] is cited for PharmAgents but points to a paper on biodegradable metal–organic frameworks, and reference [54] is cited for DrugAgent but points to a paper on lithium void formation in solid-state batteries. These citations appear to be unrelated to the claimed agents; please verify and correct.
  3. [§5, Case 1] There is a typo: "evaluate the Initiation core" should presumably read "evaluate the Initiation score." Also, "Molecular AI Agent" and "Molecule AI Agent" are used interchangeably; please standardize.
  4. [Table 1] The table lists "Biomeni" while the text and references use "Biomni." Please harmonize.
  5. [Reproducibility] Code and data processing scripts for the Case 1 snRNA-seq analysis are not provided. Given the emphasis on provenance and auditability in Section 3, the authors should make these available, ideally alongside the Supplementary Table 1 repository.

Circularity Check

1 steps flagged

Initiation Score validation is self-confirming: high-initiation cells are defined by EMT/stemness gene sets, then those same programs are reported as enrichment.

specific steps
  1. self definitional [Section 5, Case 1 (Initiation Score computation; around Figure 11C-D, after snRNA-seq analysis)]
    "Marker gene sets representing epithelial and mesenchymal states, EMT drivers, stemness regulators, glycolysis, and oxidative phosphorylation (OXPHOS) were curated. ... Cells in the top 25% (threshold = 0.324) were classified as high initiation potential, yielding 1,625 cells. ... Differential expression analysis revealed significant upregulation of CD44, FN1, and VIM, consistent with invasive phenotypes. Pathway enrichment confirmed activation of EMT transcription factors and stemness regulators in high-initiation cells."

    The high-initiation label is constructed by scoring cells with curated EMT, stemness, glycolysis/OXPHOS gene sets and then taking the top PCA-projected score. The subsequent 'confirmation' via differential expression and pathway enrichment reports EMT transcription factors and stemness regulators in cells already selected for high EMT/stemness scores. The enrichment is entailed by the selection rule, not independent evidence of invasive potential. No external endpoint (e.g., metastasis outcome, survival, distant colonization) is used, so the validation step reduces to the input gene programs by construction.

full rationale

The paper is primarily a framework vision with no formal derivation, so the central 'full-body prediction' claim is not circular—it is simply uninstantiated. The temporal-alignment difficulty admitted in Section 7 is a feasibility limitation, not circularity. The paper's many self-citations (e.g., CCPE, scIGANs, LVPT, DrugFormer) are tool references and are not used as load-bearing uniqueness theorems. The one genuine circular step is in the worked Initiation Score example: cells are classified as 'high initiation' using curated EMT/stemness/metabolism signatures, and then the same signatures are reported as enrichment/activation in those very cells. This is a self-definitional validation, not a prediction. It undermines the illustrative quantitative claim but does not by itself force the entire framework's conclusion, hence a score of 6 rather than higher.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 3 invented entities

The central claim rests on several untested domain assumptions. The framework's success depends on LLM reliability, the sufficiency of the seven-level hierarchy, the convergence of bidirectional reasoning, and the solvability of multi-modal data integration and temporal alignment. None of these are derived or benchmarked; the paper itself flags temporal alignment as unsolved. The free parameters listed belong to the illustrative Initiation Score, which uses hand-set thresholds and sample-fitted PCA weights.

free parameters (3)
  • hybrid EMT classification threshold = top 20% of normalized epithelial/mesenchymal score minimum
    Chosen in Section 5 Case 1 to define hybrid EMT cells; no biological justification or sensitivity analysis is provided.
  • high-initiation threshold = top 25% of malignant cells, score > 0.324
    Used to dichotomize cells for downstream differential expression in Section 5 Case 1; the cutoff is arbitrary within the sample.
  • PCA first-component weights = derived from first principal component of curated feature scores
    Used to aggregate molecular features into the Initiation Score; weights are fit to the same sample that is then interpreted as validation.
axioms (4)
  • domain assumption The seven biological levels (molecule, organelle, cell, tissue, organ, organ system, body) are an appropriate and complete decomposition for modeling human biology.
    The framework's task decomposition in Section 3.2 assigns all sub-tasks to these levels; if a causally important scale is missing or the boundaries are wrong, cross-scale mapping fails.
  • ad hoc to paper LLM-based agents can reliably perceive, reason, and execute domain-specific biological analyses at each scale with sufficient accuracy for cross-scale inference.
    Section 3 asserts LLMs run the whole reasoning loop (Figure 3C), but no benchmark evidence is provided for the end-to-end reliability of the proposed agent chain.
  • ad hoc to paper Bidirectional cross-scale constraint propagation converges to physiologically plausible whole-body models rather than amplifying local errors.
    Section 3.5 and Discussion claim iterative exchange until a 'convergent, physiologically plausible' model emerges; convergence is asserted, not shown.
  • domain assumption Heterogeneous multi-modal data across levels can be standardized, temporally aligned, and integrated without losing causal information.
    The Data Commons section (Section 2) presupposes this; Section 7 admits temporal alignment is 'still a largely unsolved problem.'
invented entities (3)
  • Full-Body AI Agent (supervisor) and seven level-specific basic AI agents no independent evidence
    purpose: Coordinate cross-scale reasoning and task decomposition across biological levels
    Purely conceptual; no implementation, API, or benchmark gives an independent falsifiable handle.
  • Metastasis AI Agent with Initiation, Dissemination, and Colonization scores no independent evidence
    purpose: Quantify metastatic potential in three phases from multi-scale features
    Only the Initiation score was sketched on one sample; no external validation or predicted outcomes are provided.
  • Drug AI Agent no independent evidence
    purpose: Integrate full-body physiological constraints into drug development
    Described as a design; no system, outputs, or prospective validation are shown.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Multi-Layered Framework for Modeling Human Biology: From Basic AI Agents to a Full-Body AI Agent." pith.science (2026). https://pith.science/paper/K4KIHDVD

@misc{pith2026250819800,
  author       = {Pith},
  title        = {Pith review of: A Multi-Layered Framework for Modeling Human Biology: From Basic AI Agents to a Full-Body AI Agent},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K4KIHDVD}},
  note         = {Machine review of arXiv:2508.19800}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We envision the Full-Body AI Agent as a comprehensive AI system designed to simulate, analyze, and optimize the dynamic processes of the human body across multiple biological levels. By integrating computational models, machine learning tools, and experimental platforms, this system aims to replicate and predict both physiological and pathological processes, ranging from molecules and cells to tissues, organs, and entire body systems. Central to the Full-Body AI Agent is its emphasis on integration and coordination across these biological levels, enabling analysis of how molecular changes influence cellular behaviors, tissue responses, organ function, and systemic outcomes. With a focus on biological functionality, the system is designed to advance the understanding of disease mechanisms, support the development of therapeutic interventions, and enhance personalized medicine. We propose two specialized implementations to demonstrate the utility of this framework: (1) the metastasis AI Agent, a multi-scale metastasis scoring system that characterizes tumor progression across the initiation, dissemination, and colonization phases by integrating molecular, cellular, and systemic signals; and (2) the drug AI Agent, a system-level drug development paradigm in which a drug AI-Agent dynamically guides preclinical evaluations, including organoids and chip-based models, by providing full-body physiological constraints. This approach enables the predictive modeling of long-term efficacy and toxicity beyond what localized models alone can achieve. These two agents illustrate the potential of Full-Body AI Agent to address complex biomedical challenges through multi-level integration and cross-scale reasoning.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

136 extracted references · 74 canonical work pages

  1. [1]

    Mapping the multiscale structure of biological systems,

    L. V. Schaffer and T. Ideker, "Mapping the multiscale structure of biological systems," Cell Systems, vol. 12, no. 6, pp. 622-635, 2021/06/16/ 2021, doi: https://doi.org/10.1016/j.cels.2021.05.012

    work page doi:10.1016/j.cels.2021.05.012 2021

  2. [2]

    Systems Biology: A Brief Overview,

    H. Kitano, "Systems Biology: A Brief Overview," Science (New York, N.Y.), vol. 295, pp. 1662-4, 04/01 2002, doi: 10.1126/science.1069492

    doi:10.1126/science.1069492 2002

  3. [3]

    Chapter 6 - Artificial intelligence in systems biology,

    A. Dasgupta and R. K. De, "Chapter 6 - Artificial intelligence in systems biology," in Handbook of Statistics, vol. 49, S. G. Krantz, A. S. R. Srinivasa Rao, and C. R. Rao Eds.: Elsevier, 2023, pp. 153-201

    2023

  4. [4]

    Artificial Intelligence in Bioinformatics: Advancements and Applications,

    A. Ethan and E. Edward, "Artificial Intelligence in Bioinformatics: Advancements and Applications," 08/07 2023

    2023

  5. [5]

    safe and effective

    Applications of the Full-body AI-Agent System: Case Studies The Full-body AI-Agent framework enables a systematic, multi-scale approach to dissecting disease mechanismsandoptimizingtherapeuticinterventions.Here,wepresenttwocasestudies:theexplorationof Lung cancer metastasis (Figure 11) and Drug development (Figure 12), demonstrating the framework’s potent...

  6. [6]

    black boxes,

    Current Progress and Challenges By integrating artificial intelligence across multiple biological scales, the Full-Body AI Agent framework represents a paradigm shift in how human biology is modeled and interpreted. While recent advances demonstrate its promise, several key challenges remain. These challenges encompass existing data availability,technical...

  7. [7]

    black box

    Discussion The Full-Body AI-Agent framework, as a forward-thinking proposal, presents an innovative approach to modeling human biology through a network of specialized AI Agents. Designed to operate at multiple biologicallevels,thisframeworkoffersanovellensthroughwhichtosimulateandpredictcomplexbiological processes. This framework envisions an unprecedent...

    arXiv

  8. [8]

    A deep-learning framework to predict cancer treatment response from histopathology images through imputed transcriptomics,

    D.-T. Hoang et al., "A deep-learning framework to predict cancer treatment response from histopathology images through imputed transcriptomics," Nature Cancer, vol. 5, no. 9, pp. 1305-1317, 2024/09/01 2024, doi: 10.1038/s43018-024-00793-2

    doi:10.1038/s43018-024-00793-2 2024

  9. [9]

    Designing deep learning studies in cancer diagnostics,

    A. Kleppe, O.-J. Skrede, S. De Raedt, K. Liestøl, D. J. Kerr, and H. E. Danielsen, "Designing deep learning studies in cancer diagnostics," Nature Reviews Cancer, vol. 21, no. 3, pp. 199-211, 2021/03/01 2021, doi: 10.1038/s41568-020-00327-9

    work page doi:10.1038/s41568-020-00327-9 2021

  10. [10]

    Biology-guided deep learning predicts prognosis and cancer immunotherapy response,

    Y. Jiang et al., "Biology-guided deep learning predicts prognosis and cancer immunotherapy response," Nature Communications, vol. 14, no. 1, p. 5135, 2023/08/23 2023, doi: 10.1038/s41467-023-40890-x

    work page doi:10.1038/s41467-023-40890-x 2023

  11. [11]

    Clinical and genetic associations of deep learning-derived cardiac magnetic resonance-based left ventricular mass,

    S. Khurshid et al., "Clinical and genetic associations of deep learning-derived cardiac magnetic resonance-based left ventricular mass," Nature Communications, vol. 14, no. 1, p. 1558, 2023/03/21 2023, doi: 10.1038/s41467-023-37173-w

    work page doi:10.1038/s41467-023-37173-w 2023

  12. [12]

    Predicting the clinical impact of human mutation with deep neural networks,

    L. Sundaram et al., "Predicting the clinical impact of human mutation with deep neural networks," Nature Genetics, vol. 50, no. 8, pp. 1161-1170, 2018/08/01 2018, doi: 10.1038/s41588-018-0167-z

    work page doi:10.1038/s41588-018-0167-z 2018

  13. [13]

    Unlocking precision medicine: clinical applications of integrating health records, genetics, and immunology through artificial intelligence,

    Y.-M. Chen, T.-H. Hsiao, C.-H. Lin, and Y. C. Fann, "Unlocking precision medicine: clinical applications of integrating health records, genetics, and immunology through artificial intelligence," Journal of Biomedical Science, vol. 32, no. 1, p. 16, 2025/02/07 2025, doi: 10.1186/s12929-024-01110-w

    work page doi:10.1186/s12929-024-01110-w 2025

  14. [14]

    Deep learning in pharmacogenomics: from gene regulation to patient stratification,

    A. A. Kalinin et al., "Deep learning in pharmacogenomics: from gene regulation to patient stratification," (in eng), Pharmacogenomics, vol. 19, no. 7, pp. 629-650, May 2018, doi: 10.2217/pgs-2018-0008

    work page doi:10.2217/pgs-2018-0008 2018

  15. [15]

    Prediction of drug efficacy from transcriptional profiles with deep learning,

    J. Zhu et al., "Prediction of drug efficacy from transcriptional profiles with deep learning," Nature Biotechnology, vol. 39, no. 11, pp. 1444-1452, 2021/11/01 2021, doi: 10.1038/s41587-021-00946-z

    work page doi:10.1038/s41587-021-00946-z 2021

  16. [16]

    Artificial intelligence for modelling infectious disease epidemics,

    M. U. G. Kraemer et al., "Artificial intelligence for modelling infectious disease epidemics," Nature, vol. 638, no. 8051, pp. 623-635, 2025/02/01 2025, doi: 10.1038/s41586-024-08564-w

    work page doi:10.1038/s41586-024-08564-w 2025

  17. [17]

    Machine learning for data-centric epidemic forecasting,

    A. Rodríguez et al., "Machine learning for data-centric epidemic forecasting," Nature Machine Intelligence, vol. 6, no. 10, pp. 1122-1131, 2024/10/01 2024, doi: 10.1038/s42256-024-00895-7

    work page doi:10.1038/s42256-024-00895-7 2024

  18. [18]

    Harnessing machine learning to guide phylogenetic-tree search algorithms,

    D. Azouri, S. Abadi, Y. Mansour, I. Mayrose, and T. Pupko, "Harnessing machine learning to guide phylogenetic-tree search algorithms," Nature Communications, vol. 12, no. 1, p. 1983, 2021/03/31 2021, doi: 10.1038/s41467-021-22073-8

    doi:10.1038/s41467-021-22073-8 1983

  19. [19]

    ModelRevelator: Fast phylogenetic model estimation via deep learning,

    S. Burgstaller-Muehlbacher, S. M. Crotty, H. A. Schmidt, F. Reden, T. Drucks, and A. von Haeseler, "ModelRevelator: Fast phylogenetic model estimation via deep learning," Molecular Phylogenetics and Evolution, vol. 188, p. 107905, 2023/11/01/ 2023, doi: https://doi.org/10.1016/j.ympev.2023.107905

    arXiv 2023

  20. [20]

    Genome modeling and design across all domains of life with Evo 2,

    G. Brixi et al., "Genome modeling and design across all domains of life with Evo 2," bioRxiv, p. 2025.02.18.638918, 2025, doi: 10.1101/2025.02.18.638918

    doi:10.1101/2025.02.18.638918 2025

  21. [21]

    Perspectives on computational modeling of biological systems and the significance of the SysMod community,

    B. L. Puniya, M. Verma, C. Damiani, S. Bakr, and A. Dräger, "Perspectives on computational modeling of biological systems and the significance of the SysMod community," (in eng), Bioinform Adv, vol. 4, no. 1, p. vbae090, 2024, doi: 10.1093/bioadv/vbae090. 47

    work page doi:10.1093/bioadv/vbae090 2024

  22. [22]

    AI-driven multi-omics integration for multi-scale predictive modeling of genotype-environment-phenotype relationships,

    Y. Wu and L. Xie, "AI-driven multi-omics integration for multi-scale predictive modeling of genotype-environment-phenotype relationships," Computational and Structural Biotechnology Journal, vol. 27, pp. 265-277, 2025/01/01/ 2025, doi: https://doi.org/10.1016/j.csbj.2024.12.030

    work page doi:10.1016/j.csbj.2024.12.030 2025

  23. [23]

    A multiscale approach for biomedical machine learning,

    "A multiscale approach for biomedical machine learning," Nature Machine Intelligence, vol. 6, no. 9, pp. 989-989, 2024/09/01 2024, doi: 10.1038/s42256-024-00907-6

    work page doi:10.1038/s42256-024-00907-6 2024

  24. [24]

    INSIHGT: an accessible multi-scale, multi-modal 3D spatial biology platform,

    C. N. Yau et al., "INSIHGT: an accessible multi-scale, multi-modal 3D spatial biology platform," Nature Communications, vol. 15, no. 1, p. 10888, 2024/12/30 2024, doi: 10.1038/s41467-024-55248-0

    work page doi:10.1038/s41467-024-55248-0 2024

  25. [25]

    Opportunities and challenges in using real-world data for health care,

    V. A. Rudrapatna and A. J. Butte, "Opportunities and challenges in using real-world data for health care," (in eng), J Clin Invest, vol. 130, no. 2, pp. 565-574, Feb 3 2020, doi: 10.1172/jci129197

    work page doi:10.1172/jci129197 2020

  26. [26]

    Autonomous chemical research with large language models,

    D. A. Boiko, R. MacKnight, B. Kline, and G. Gomes, "Autonomous chemical research with large language models," Nature, vol. 624, no. 7992, pp. 570-578, 2023/12/01 2023, doi: 10.1038/s41586-023-06792-0

    doi:10.1038/s41586-023-06792-0 2023

  27. [27]

    Language models for biological research: a primer,

    E. Simon, K. Swanson, and J. Zou, "Language models for biological research: a primer," (in eng), Nat Methods, vol. 21, no. 8, pp. 1422-1429, Aug 2024, doi: 10.1038/s41592-024-02354-y

    work page doi:10.1038/s41592-024-02354-y 2024

  28. [28]

    Empowering biomedical discovery with AI agents,

    S. Gao et al., "Empowering biomedical discovery with AI agents," Cell, vol. 187, no. 22, pp. 6125-6151, 2024, doi: 10.1016/j.cell.2024.09.022

    doi:10.1016/j.cell.2024.09.022 2024

  29. [29]

    Large language models streamline automated machine learning for clinical studies,

    S. Tayebi Arasteh et al., "Large language models streamline automated machine learning for clinical studies," Nature Communications, vol. 15, no. 1, p. 1603, 2024/02/21 2024, doi: 10.1038/s41467-024-45879-8

    work page doi:10.1038/s41467-024-45879-8 2024

  30. [30]

    A pathologist–AI collaboration framework for enhancing diagnostic accuracies and efficiencies,

    Z. Huang et al., "A pathologist–AI collaboration framework for enhancing diagnostic accuracies and efficiencies," Nature Biomedical Engineering, vol. 9, no. 4, pp. 455-470, 2025/04/01 2025, doi: 10.1038/s41551-024-01223-5

    work page doi:10.1038/s41551-024-01223-5 2025

  31. [31]

    SpatialAgent: An autonomous AI agent for spatial biology,

    H. Wang et al., "SpatialAgent: An autonomous AI agent for spatial biology," bioRxiv, p. 2025.04.03.646459, 2025, doi: 10.1101/2025.04.03.646459

    doi:10.1101/2025.04.03.646459 2025

  32. [32]

    How to build the virtual cell with artificial intelligence: Priorities and opportunities,

    C. Bunne et al., "How to build the virtual cell with artificial intelligence: Priorities and opportunities," Cell, vol. 187, no. 25, pp. 7045-7063, 2024, doi: 10.1016/j.cell.2024.11.015

    doi:10.1016/j.cell.2024.11.015 2024

  33. [33]

    Robin: A multi-agent system for automating scientific discovery,

    A. Essam Ghareeb et al., "Robin: A multi-agent system for automating scientific discovery," p. arXiv:2505.13400doi: 10.48550/arXiv.2505.13400

    Pith/arXiv arXiv doi:10.48550/arxiv.2505.13400

  34. [34]

    OriGene: A Self-Evolving Virtual Disease Biologist Automating Therapeutic Target Discovery,

    Z. Zhang et al., "OriGene: A Self-Evolving Virtual Disease Biologist Automating Therapeutic Target Discovery," bioRxiv, p. 2025.06.03.657658, 2025, doi: 10.1101/2025.06.03.657658

    doi:10.1101/2025.06.03.657658 2025

  35. [35]

    Biomni: A General-Purpose Biomedical AI Agent,

    K. Huang et al., "Biomni: A General-Purpose Biomedical AI Agent," bioRxiv, p. 2025.05.30.656746, 2025, doi: 10.1101/2025.05.30.656746

    doi:10.1101/2025.05.30.656746 2025

  36. [36]

    Enhancing diagnostic capability with multi-agents conversational large language models,

    X. Chen et al., "Enhancing diagnostic capability with multi-agents conversational large language models," (in eng), NPJ Digit Med, vol. 8, no. 1, p. 159, Mar 13 2025, doi: 10.1038/s41746-025-01550-0

    doi:10.1038/s41746-025-01550-0 2025

  37. [37]

    LLM Agent Swarm for Hypothesis-Driven Drug Discovery,

    K. Song, A. Trotter, and J. Y. Chen, "LLM Agent Swarm for Hypothesis-Driven Drug Discovery," 2025, 2025

    2025

  38. [38]

    Towards an AI co-scientist,

    J. Gottweis et al., "Towards an AI co-scientist," 2025, 2025

    2025

  39. [39]

    Big data: Historic advances and emerging trends in biomedical research,

    C. J. Cremin, S. Dash, and X. Huang, "Big data: Historic advances and emerging trends in biomedical research," Current Research in Biotechnology, vol. 4, pp. 138-151, 2022/01/01/ 2022, doi: https://doi.org/10.1016/j.crbiot.2022.02.004

    work page doi:10.1016/j.crbiot.2022.02.004 2022

  40. [40]

    Biomedical Big Data Technologies, Applications, and Challenges for Precision Medicine: A Review,

    X. Yang, K. Huang, D. Yang, W. Zhao, and X. Zhou, "Biomedical Big Data Technologies, Applications, and Challenges for Precision Medicine: A Review," (in eng), Glob Chall, vol. 8, no. 1, p. 2300163, Jan 2024, doi: 10.1002/gch2.202300163

    work page doi:10.1002/gch2.202300163 2024

  41. [41]

    Integrating and formatting biomedical data as pre-calculated knowledge graph embeddings in the Bioteque,

    A. Fernández-Torras, M. Duran-Frigola, M. Bertoni, M. Locatelli, and P. Aloy, "Integrating and formatting biomedical data as pre-calculated knowledge graph embeddings in the Bioteque," Nature Communications, vol. 13, no. 1, p. 5304, 2022/09/09 2022, doi: 10.1038/s41467-022-33026-0

    doi:10.1038/s41467-022-33026-0 2022

  42. [42]

    Sequence modeling and design from molecular to genome scale with Evo,

    E. Nguyen et al., "Sequence modeling and design from molecular to genome scale with Evo," Science, vol. 386, no. 6723, p. eado9336, 2024, doi: doi:10.1126/science.ado9336

    doi:10.1126/science.ado9336 2024

  43. [43]

    BioLLM: A standardized framework for integrating and benchmarking single-cell foundation models,

    P. Qiu et al., "BioLLM: A standardized framework for integrating and benchmarking single-cell foundation models," Patterns, vol. 6, no. 8, 2025, doi: 10.1016/j.patter.2025.101326

    arXiv 2025

  44. [44]

    scGPT: toward building a foundation model for single-cell multi-omics using generative AI,

    H. Cui et al., "scGPT: toward building a foundation model for single-cell multi-omics using generative AI," Nat Methods, vol. 21, no. 8, pp. 1470-1480, 2024/08/01 2024, doi: 10.1038/s41592-024-02201-0. 48

    doi:10.1038/s41592-024-02201-0 2024

  45. [45]

    Intelligent single-cell manipulation: LLMs- and object detection-enhanced active-matrix digital microfluidics,

    Z. Jia et al., "Intelligent single-cell manipulation: LLMs- and object detection-enhanced active-matrix digital microfluidics," Microsystems & Nanoengineering, vol. 11, no. 1, p. 133, 2025/07/07 2025, doi: 10.1038/s41378-025-00962-y

    work page doi:10.1038/s41378-025-00962-y 2025

  46. [46]

    SpaCCC: Large language model-based cell-cell communication inference for spatially resolved transcriptomic data,

    B. Ji, L. Xu, and S. Peng, "SpaCCC: Large language model-based cell-cell communication inference for spatially resolved transcriptomic data," bioRxiv, p. 2024.02.21.581369, 2024, doi: 10.1101/2024.02.21.581369

    work page doi:10.1101/2024.02.21.581369 2024

  47. [47]

    QuST-LLM: Integrating Large Language Models for Comprehensive Spatial Transcriptomics Analysis,

    C. H. Huang, "QuST-LLM: Integrating Large Language Models for Comprehensive Spatial Transcriptomics Analysis," arXiv, preprint 2024. [Online]. Available: https://arxiv.org/abs/2406.14307

    Pith/arXiv arXiv 2024

  48. [48]

    A vision–language foundation model for precision oncology,

    J. Xiang et al., "A vision–language foundation model for precision oncology," Nature, vol. 638, no. 8051, pp. 769-778, 2025/02/01 2025, doi: 10.1038/s41586-024-08378-w

    doi:10.1038/s41586-024-08378-w 2025

  49. [49]

    Large Language Models in radiology: A technical and clinical perspective,

    J.-P. Kao and H.-T. Kao, "Large Language Models in radiology: A technical and clinical perspective," European Journal of Radiology Artificial Intelligence, vol. 2, p. 100021, 2025/06/01/ 2025, doi: https://doi.org/10.1016/j.ejrai.2025.100021

    arXiv 2025

  50. [50]

    LLM-MedQA: Enhancing Medical Question Answering through Case Studies in Large Language Models,

    H. Yang et al., "LLM-MedQA: Enhancing Medical Question Answering through Case Studies in Large Language Models," p. arXiv:2501.05464doi: 10.48550/arXiv.2501.05464

    Pith/arXiv arXiv doi:10.48550/arxiv.2501.05464

  51. [51]

    The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery,

    C. Lu, C. Lu, R. T. Lange, J. Foerster, J. Clune, and D. Ha, "The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery," arXiv, arXiv:2408.06292 cs.AI 2024. [Online]. Available: https://arxiv.org/abs/2408.06292

    Pith/arXiv arXiv 2024

  52. [52]

    Spatialagent: An autonomous ai agent for spatial biology,

    "Spatialagent: An autonomous ai agent for spatial biology," bioRxiv, pp. 2025–04, 2025, doi: 10.1101/2025.04.03.646459v1

    work page doi:10.1101/2025.04.03.646459v1 2025

  53. [53]

    TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools,

    S. Gao et al., "TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools," arXiv, arXiv:2503.10970 cs.AI 2025. [Online]. Available: https://arxiv.org/abs/2503.10970

    Pith/arXiv arXiv 2025

  54. [54]

    ResearchAgent: Iterative Research Idea Generation over Scientific Literature with Large Language Models,

    J. Baek, S. K. Jauhar, S. Cucerzan, and S. J. Hwang, "ResearchAgent: Iterative Research Idea Generation over Scientific Literature with Large Language Models," arXiv preprint, 10.48550/arXiv.2404.07738 2025. [Online]. Available: https://arxiv.org/abs/2404.07738

    Pith/arXiv arXiv doi:10.48550/arxiv.2404.07738 2025

  55. [55]

    ReAct: Synergizing Reasoning and Acting in Language Models,

    S. Yao et al., "ReAct: Synergizing Reasoning and Acting in Language Models," arXiv preprint, 10.48550/arXiv.2210.03629 2023. [Online]. Available: https://arxiv.org/abs/2210.03629

    Pith/arXiv arXiv doi:10.48550/arxiv.2210.03629 2023

  56. [56]

    Bridging biodegradable metals and biodegradable polymers: A comprehensive review of biodegradable metal–organic frameworks for biomedical application,

    T. Zhang et al., "Bridging biodegradable metals and biodegradable polymers: A comprehensive review of biodegradable metal–organic frameworks for biomedical application," Progress in Materials Science, vol. 155, p. 101526, 2026/01/01/ 2026, doi: https://doi.org/10.1016/j.pmatsci.2025.101526

    arXiv 2026

  57. [57]

    Mitigating lithium void formation in all-solid-state batteries via a high lithium diffusion 3D composite interlayer,

    Z.-N. Cheng et al., "Mitigating lithium void formation in all-solid-state batteries via a high lithium diffusion 3D composite interlayer," Journal of Energy Chemistry, vol. 111, pp. 86-93, 2025/12/01/ 2025, doi: https://doi.org/10.1016/j.jechem.2025.07.033

    work page doi:10.1016/j.jechem.2025.07.033 2025

  58. [58]

    Fleming: An AI Agent for Antibiotic Discovery in Mycobacterium Tuberculosis,

    Z. Wei et al., "Fleming: An AI Agent for Antibiotic Discovery in Mycobacterium Tuberculosis," bioRxiv, p. 2025.04, 2025, doi: 10.1101/2025.04.01.646719v1

    work page doi:10.1101/2025.04.01.646719v1 2025

  59. [59]

    Tx-LLM: A Large Language Model for Therapeutics,

    J. M. Zambrano Chaves et al., "Tx-LLM: A Large Language Model for Therapeutics," arXiv preprint, 10.48550/arXiv.2406.06316 2024. [Online]. Available: https://arxiv.org/abs/2406.06316

    Pith/arXiv arXiv doi:10.48550/arxiv.2406.06316 2024

  60. [60]

    TxGemma: Efficient and Agentic LLMs for Therapeutics,

    E. Wang et al., "TxGemma: Efficient and Agentic LLMs for Therapeutics," arXiv preprint, arXiv:2504.06196 cs.AI 2025. [Online]. Available: https://arxiv.org/abs/2504.06196

    Pith/arXiv arXiv 2025

  61. [61]

    ClinicalAgent: Clinical Trial Multi-Agent System with Large Language Model-based Reasoning,

    L. Yue, S. Xing, J. Chen, and T. Fu, "ClinicalAgent: Clinical Trial Multi-Agent System with Large Language Model-based Reasoning," arXiv, preprint 2024. [Online]. Available: https://arxiv.org/abs/2404.14777

    Pith/arXiv arXiv 2024

  62. [62]

    Automated Hypothesis Validation with Agentic Sequential Falsifications,

    K. Huang, Y. Jin, R. Li, M. Y. Li, E. Candès, and J. Leskovec, "Automated Hypothesis Validation with Agentic Sequential Falsifications," arXiv, arXiv:2502.09858 cs.LG 2025. [Online]. Available: https://arxiv.org/abs/2502.09858

    Pith/arXiv arXiv 2025

  63. [63]

    Autonomous LLM-driven research from data to human-verifiable research papers,

    T. Ifargan, L. Hafner, M. Kern, O. Alcalay, and R. Kishony, "Autonomous LLM-driven research from data to human-verifiable research papers," arXiv preprint, 10.48550/arXiv.2404.17605 2024. [Online]. Available: https://arxiv.org/abs/2404.17605. 49

    Pith/arXiv arXiv doi:10.48550/arxiv.2404.17605 2024

  64. [64]

    Polygenic enrichment distinguishes disease associations of individual cells in single-cell RNA-seq data,

    M. J. Zhang et al., "Polygenic enrichment distinguishes disease associations of individual cells in single-cell RNA-seq data," Nature Genetics, vol. 54, no. 10, pp. 1572-1580, 2022/10/01 2022, doi: 10.1038/s41588-022-01167-z

    work page doi:10.1038/s41588-022-01167-z 2022

  65. [65]

    The Interactome of Protein, DNA, and RNA,

    Y. Yan and T. Huang, "The Interactome of Protein, DNA, and RNA," (in eng), Methods Mol Biol, vol. 2695, pp. 89-110, 2023, doi: 10.1007/978-1-0716-3346-5_6

    work page doi:10.1007/978-1-0716-3346-5_6 2023

  66. [66]

    Understanding Protein Functions in the Biological Context,

    W. Zhang and T. Wang, "Understanding Protein Functions in the Biological Context," (in eng), Protein Pept Lett, vol. 30, no. 6, pp. 449-458, 2023, doi: 10.2174/0929866530666230507212638

    work page doi:10.2174/0929866530666230507212638 2023

  67. [67]

    Advancing bioinformatics with large language models: components, applications and perspectives,

    J. Liu et al., "Advancing bioinformatics with large language models: components, applications and perspectives," (in eng), ArXiv, Jan 31 2025

    2025

  68. [68]

    Mitochondrial OXPHOS complex assembly lines,

    L. E. Formosa and M. T. Ryan, "Mitochondrial OXPHOS complex assembly lines," Nature Cell Biology, vol. 20, no. 5, pp. 511-513, 2018/05/01 2018, doi: 10.1038/s41556-018-0098-z

    work page doi:10.1038/s41556-018-0098-z 2018

  69. [69]

    Mitochondrial dysfunction in neurodegenerative disorders,

    M. M. Klemmensen, S. H. Borrowman, C. Pearce, B. Pyles, and B. Chandra, "Mitochondrial dysfunction in neurodegenerative disorders," (in eng), Neurotherapeutics, vol. 21, no. 1, p. e00292, Jan 2024, doi: 10.1016/j.neurot.2023.10.002

    work page doi:10.1016/j.neurot.2023.10.002 2024

  70. [70]

    Mitochondrial DNA variation and cancer,

    P. K. Kopinski, L. N. Singh, S. Zhang, M. T. Lott, and D. C. Wallace, "Mitochondrial DNA variation and cancer," Nature Reviews Cancer, vol. 21, no. 7, pp. 431-445, 2021/07/01 2021, doi: 10.1038/s41568-021-00358-w

    work page doi:10.1038/s41568-021-00358-w 2021

  71. [71]

    Mitochondrial DNA mutation enhances sensitivity to immunotherapy in melanoma,

    "Mitochondrial DNA mutation enhances sensitivity to immunotherapy in melanoma," Nature Cancer, vol. 5, no. 4, pp. 544-545, 2024/04/01 2024, doi: 10.1038/s43018-023-00722-9

    work page doi:10.1038/s43018-023-00722-9 2024

  72. [72]

    Mitochondrial and metabolic dysfunction in ageing and age-related diseases,

    J. A. Amorim, G. Coppotelli, A. P. Rolo, C. M. Palmeira, J. M. Ross, and D. A. Sinclair, "Mitochondrial and metabolic dysfunction in ageing and age-related diseases," Nature Reviews Endocrinology, vol. 18, no. 4, pp. 243-258, 2022/04/01 2022, doi: 10.1038/s41574-021-00626-7

    work page doi:10.1038/s41574-021-00626-7 2022

  73. [73]

    Mitochondrial DNA-triggered innate immune response: mechanisms and diseases,

    M.-M. Hu and H.-B. Shu, "Mitochondrial DNA-triggered innate immune response: mechanisms and diseases," Cellular & Molecular Immunology, vol. 20, no. 12, pp. 1403-1412, 2023/12/01 2023, doi: 10.1038/s41423-023-01086-x

    work page doi:10.1038/s41423-023-01086-x 2023

  74. [74]

    Multifaceted mitochondria: moving mitochondrial science beyond function and dysfunction,

    A. S. Monzel, J. A. Enríquez, and M. Picard, "Multifaceted mitochondria: moving mitochondrial science beyond function and dysfunction," Nature Metabolism, vol. 5, no. 4, pp. 546-562, 2023/04/01 2023, doi: 10.1038/s42255-023-00783-1

    work page doi:10.1038/s42255-023-00783-1 2023

  75. [75]

    DeepMito: accurate prediction of protein sub-mitochondrial localization using convolutional neural networks,

    C. Savojardo, N. Bruciaferri, G. Tartari, P. L. Martelli, and R. Casadio, "DeepMito: accurate prediction of protein sub-mitochondrial localization using convolutional neural networks," (in eng), Bioinformatics, vol. 36, no. 1, pp. 56-64, Jan 1 2020, doi: 10.1093/bioinformatics/btz512

    work page doi:10.1093/bioinformatics/btz512 2020

  76. [76]

    MITOMAP: a human mitochondrial genome database,

    A. M. Kogelnik, M. T. Lott, M. D. Brown, S. B. Navathe, and D. C. Wallace, "MITOMAP: a human mitochondrial genome database," (in eng), Nucleic Acids Res, vol. 24, no. 1, pp. 177-9, Jan 1 1996, doi: 10.1093/nar/24.1.177

    work page doi:10.1093/nar/24.1.177 1996

  77. [77]

    MitoTrace: A Computational Framework for Analyzing Mitochondrial Variation in Single-Cell RNA Sequencing Data,

    M. Wang, W. Deng, D. C. Samuels, Z. Zhao, and L. M. Simon, "MitoTrace: A Computational Framework for Analyzing Mitochondrial Variation in Single-Cell RNA Sequencing Data," (in eng), Genes (Basel), vol. 14, no. 6, Jun 4 2023, doi: 10.3390/genes14061222

    work page doi:10.3390/genes14061222 2023

  78. [78]

    CellMarker: a manually curated resource of cell markers in human and mouse,

    X. Zhang et al., "CellMarker: a manually curated resource of cell markers in human and mouse," (in eng), Nucleic Acids Res, vol. 47, no. D1, pp. D721-d728, Jan 8 2019, doi: 10.1093/nar/gky900

    work page doi:10.1093/nar/gky900 2019

  79. [79]

    PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data,

    O. Franzén, L. M. Gan, and J. L. M. Björkegren, "PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data," (in eng), Database (Oxford), vol. 2019, Jan 1 2019, doi: 10.1093/database/baz046

    doi:10.1093/database/baz046 2019

  80. [80]

    Reversed graph embedding resolves complex single-cell trajectories,

    X. Qiu et al., "Reversed graph embedding resolves complex single-cell trajectories," Nat Methods, vol. 14, no. 10, pp. 979-982, 2017/10/01 2017, doi: 10.1038/nmeth.4402

    doi:10.1038/nmeth.4402 2017

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.