TCMScience: An Autonomous Scientist for Traditional Chinese Medicine

arXiv 2026 65.3 method

TLDR

TCMScience is an autonomous TCM knowledge-discovery agent that separates language-model proposals from execution, uses claim-support contracts, and evaluates network-pharmacology workflows.

Reasoning

The paper introduces a principled architecture for autonomous scientific discovery with extensive resource catalogs and a concrete case study, which are strengths. However, the abstract is truncated and lacks full quantitative results and explicit limitations, making the empirical evidence incomplete.

Read-first score

Read-first score 65.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 72.

Recency 8%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=ablation,analysis,dataset,evaluation,result

Topical relevance 42%

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=dataset

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 25.

Keyword Scores

autonomous research agent

automated scientific discovery

AI for scientific research

scientific discovery agent

AI scientist

research automation

automated research

automated experimentation

experiment design agent

literature review agent

survey generation

paper writing agent

Deep Analysis

Innovations

Separation of language-model proposals from execution and release authority
Architecture combining a trusted kernel, a scientific compiler, and claim-support contracts
Compilation of plans into typed operations with explicit execution permissions and pre-release checks for source provenance, study design, and claim scope
Large model-independent resource layer: 117 public-source connectors with 444 typed operations across 29 domains, 150 native tools across 12 domains, and a TCM data hub with 136 databases, 76 dataset specifications, and 84,294 formula-source records
Layered governance that jointly reduces release of erroneous outputs

Methodology

TCMScience uses a trusted kernel, scientific compiler, and claim-support contracts to compile language-model plans into typed operations, bound execution by explicit permissions, and check candidate findings against source provenance, study design, and claim scope before release. It integrates a capability registry of 117 public-source connectors with 444 typed operations across 29 domains, 150 native tools across 12 domains, and a TCM data hub cataloging 136 databases, 76 dataset specifications, and 84,294 formula-source records, supporting browser and local execution interfaces. Evaluations include a network-pharmacology analysis of Gegen Qinlian Tang and a governance ablation over 121 constructed erroneous outputs spanning 26 classes plus 10 compliant controls.

Key Results

In network-pharmacology analysis, replacing the whole-proteome background with the assayed-protein background changed significant pathways from 66 to zero, while incorporating inactive screening records identified nine pathways including drug-metabolism signals retained as mechanistic hypotheses. In the governance ablation, 121 errors were released without governance, 57 with the domain layer, 7 with the kernel, and 0 with both layers; all 10 compliant controls were retained.

Resources

Paper