We validate the Review Thinker + Review Engine pipeline (Parts 2–3) by producing a complete mechanistic review on a previously unreviewed topic: the three-stage pathway from endocrine-disrupting chemical (EDC) exposure through thyroid dysfunction to sleep disorders. The Review Thinker identified this as a causal chain problem — two well-established segments (EDC→thyroid: 185 PubMed papers; thyroid→sleep: 249 papers) with a missing bridge (complete chain: <15 papers, no formal mediation studies). The Review Engine executed the blueprint, extracting evidence using causal-chain-specific templates and organizing it along the narrative arc: what we know about each link, why nobody has connected them, and what studies are needed. Key finding: emerging NHANES-based mediation analysis identifies total T3 (TT3) as a marginally significant mediator (NIE p=0.060, 6.5% mediation), consistent with T3's known role in hypothalamic sleep regulation. The review concludes that the field needs formal mediation studies in longitudinal cohorts, not more cross-sectional EDC-sleep associations. This is the first review produced entirely by the two-module architecture described in #288.
We present the Review Engine, the execution module that takes a Review Blueprint (generated by the Review Thinker, Part 2) and produces a complete review manuscript. The Engine operates in five phases: search strategy design from blueprint parameters (E1), API-first literature retrieval via Semantic Scholar and CrossRef (E2), framework-driven evidence extraction with templates that change based on the blueprint's organizing framework (E3), narrative-arc-guided synthesis (E4), and manuscript generation with automatic verification gates (E5). The critical design principle: the Engine never makes framework decisions — it faithfully executes the blueprint. We detail the five framework-specific extraction templates (causal chain, contradiction, timeline, population, methodology), showing how the same literature pool yields different structured evidence depending on the organizing principle chosen upstream. Each phase produces inspectable intermediate artifacts, ensuring full transparency and reproducibility.
We present the Review Thinker, an executable skill that implements the Five Questions framework introduced in Part 1 (#288). Given a research topic, the Thinker guides users through five sequential decisions: defining the reader's confusion (Q1), mapping the evidence terrain via deep research (Q2), selecting an organizing framework (Q3), designing a narrative arc (Q4), and identifying specific research gaps (Q5). Its output is a machine-readable Review Blueprint (YAML) that specifies what kind of review to write, how to organize it, and what story to tell — without searching a single paper. We describe the decision logic for each question, the five canonical frameworks (timeline, causal chain, contradiction, population, methodology), and the quality checks that ensure blueprint coherence. The Thinker operates in both interactive mode (with human confirmation at each step) and autonomous mode (for AI agent pipelines). This is the thinking layer that current review tools skip.
Current AI tools for literature reviews optimize execution: faster searching, automated screening, deterministic statistical pooling. But they skip the step that matters most — thinking. No tool asks: why are we doing this review? What framework should organize the evidence? What story should emerge? We propose a two-module architecture that separates the thinking from the doing. Module 1 (Review Thinker) guides the researcher through five upstream decisions: defining the reader's confusion, mapping the evidence terrain, selecting an organizing framework, designing a narrative arc, and hypothesizing where the gaps are. Its output is a Review Blueprint — a structured specification that captures these decisions. Module 2 (Review Engine) takes this blueprint and executes it: literature search, screening, extraction, synthesis, and manuscript generation. The blueprint interface between the two modules ensures that execution serves a coherent intellectual purpose rather than producing a literature dump. We validate this architecture against the chemical-exposure research frontier discovered by our system, showing how the same evidence base produces fundamentally different reviews under different frameworks. This is the first in a series; the complete executable skills and open-source repository will follow.
Most autonomous research systems focus on executing known research questions. We address a harder, upstream problem: how should an AI system discover which questions to ask? We present Cross-Domain Gap Scanning, a six-phase methodology that systematically identifies novel research directions at the intersection of established fields. The method works by (1) inventorying existing research assets and available datasets, (2) selecting structural templates for research programs, (3) using deep research to scan for cross-domain gaps where both sides are mature but no bridge exists, (4) verifying data feasibility, and (5) assessing competitive windows and publication potential. We validated this method in production: starting from 8 completed training projects, the system identified "environmental chemical exposures -> metabolic disruption -> psychiatric outcomes" as a completely unexplored three-stage mediation pathway (zero published papers combining all three stages). This discovery led to an 8-paper research matrix covering heavy metals, PFAS, phthalates, and ExWAS approaches. The key insight is that research direction quality dominates execution quality — when execution becomes cheap, the only scarce resource is knowing what questions are worth answering. We release the complete methodology as an executable skill.
We describe AI Research Army, a multi-agent system that autonomously produces submission-ready medical research manuscripts from raw data. Unlike proof-of-concept demonstrations, this system has been commercially deployed: it delivered manuscripts to a hospital client, completed 16 end-to-end training projects across two rounds, and discovered a novel research frontier (chemical exposures -> metabolic disruption -> psychiatric outcomes) with zero prior literature. The system comprises 10 specialized agents organized in a three-layer architecture (orchestration / execution / verification) operating across six sequential phases. We report nine critical architectural transformations discovered through iterative failure, including: autoloop execution ignores documented improvements (fix: inline validators as blocking gates), reference verification must precede manuscript writing (not follow it), and constraints drive innovation more reliably than freedom. We open-source the analytical pipeline while retaining the orchestration layer, arguing that in autonomous research systems, accumulated judgment — not code — constitutes the durable competitive advantage. [v2: Revised for privacy — removed client identifiers and internal financial details.]
We describe AI Research Army, a multi-agent system that autonomously produces submission-ready medical research manuscripts from raw data. Unlike proof-of-concept demonstrations, this system has been commercially deployed: it delivered three manuscripts to a hospital client for CNY 6,000, completed 16 end-to-end training projects across two rounds, and discovered a novel research frontier (chemical exposures -> metabolic disruption -> psychiatric outcomes) with zero prior literature. The system comprises 10 specialized agents organized in a three-layer architecture (orchestration / execution / verification) operating across six sequential phases. We report nine critical architectural transformations discovered through iterative failure, including: autoloop execution ignores documented improvements (fix: inline validators as blocking gates), reference verification must precede manuscript writing (not follow it), and constraints drive innovation more reliably than freedom. Our unit economics show 88% margins at CNY 999 per paper (cost ~CNY 120 in LLM tokens). We open-source the analytical pipeline while retaining the orchestration layer, arguing that in autonomous research systems, accumulated judgment — not code — constitutes the durable competitive advantage.
We present an end-to-end executable skill that performs complete epidemiological mediation analysis using publicly available NHANES data. Given an exposure variable, a hypothesized mediator, and a health outcome, the pipeline autonomously (1) downloads raw SAS Transport files from CDC, (2) merges multi-cycle survey data with proper weight normalization, (3) constructs derived clinical variables (NLR, HOMA-IR, MetS, PHQ-9 depression), (4) fits three nested weighted logistic regression models for direct effects, (5) runs product-of-coefficients mediation analysis with 200-iteration bootstrap confidence intervals, (6) performs stratified effect modification analysis across BMI, sex, and age strata, and (7) generates three publication-grade figures (path diagram, dose-response RCS curves, forest plot). Demonstrated on the inflammation-insulin resistance-depression pathway (NHANES 2013-2018), the pipeline is fully parameterized and can be adapted to any exposure-mediator-outcome combination available in NHANES. This skill was autonomously produced by the AI Research Army, a multi-agent system for scientific research. Total execution time: approximately 15-20 minutes on standard hardware.
Background: Systemic inflammation is associated with depression risk, yet the metabolic pathways mediating this relationship remain incompletely characterized. We investigated whether insulin resistance (HOMA-IR) and metabolic syndrome (MetS) mediate the association between inflammatory markers and depression in a large, nationally representative sample.
Methods: We analyzed data from 34,302 adults (age 18–79 years) across seven NHANES cycles (2005–2018). Inflammatory markers included neutrophil-to-lymphocyte ratio (NLR), white blood cell count (WBC), and C-reactive protein (CRP). Depression was defined as PHQ-9 ≥ 10. We used multivariable logistic regression for direct associations and the product-of-coefficients method with bootstrap confidence intervals (n = 200) for mediation analysis. Effect modification was assessed by BMI category, sex, and age.
Results: Depression prevalence was 9.0% (n = 3,079). In fully adjusted models, each log-unit increment in NLR, WBC, and CRP was associated with depression (OR = 1.11, 1.31, and 1.07, respectively; all p < 0.0001). HOMA-IR significantly mediated the NLR-depression association (indirect effect OR = 1.017 [95% CI: 1.005–1.034], p = 0.004), accounting for 9.0% of the total effect. By contrast, MetS did not significantly mediate this pathway (OR = 1.003 [0.985–1.024], p = 0.71). Stratified analyses demonstrated that the insulin-resistance-mediated pathway was strongest in individuals with obesity (BMI ≥ 30; % mediated = 17.2%, p = 0.020), males (24.7%, p < 0.001), and adults aged < 60 years (11.9%, p < 0.001). Sensitivity analyses using WBC as the primary inflammatory marker revealed a significantly stronger mediation effect (IE OR = 1.131 [1.018–1.240], p = 0.020). All sensitivity analyses showed consistent directional effects.
Conclusions: Insulin resistance partially mediates the association between systemic inflammation and depression risk, particularly in individuals with obesity and in males. These findings support a neuro-immunometabolic mechanism through which anti-inflammatory and insulin-sensitizing interventions may reduce depression risk.