Constructing Legal Knowledge Graphs and Tracing Bias in Medical Artificial Intelligence Based on Graph Neural Networks

{"In":[0],"medical":[1,55,249],"artificial":[2],"intelligence":[3],"(AI)":[4],"legal":[5,24,76,107,231],"compliance":[6,143,241],"analysis,":[7],"this":[8],"paper":[9],"addresses":[10],"the":[11,18,150,173,206,214,245],"ambiguity":[12],"in":[13,28,167],"liability":[14],"determination":[15],"caused":[16],"by":[17,146],"semantic":[19,77,170],"fragmentation":[20],"of":[21,51,75,152,160,169,182,192,248],"multi-source":[22],"heterogeneous":[23,52],"texts":[25],"and":[26,38,59,63,65,82,94,126,149,186,213,233,239],"difficulties":[27],"tracing":[29],"algorithmic":[30,234],"biases.":[31],"It":[32],"constructs":[33],"a":[34,66,85,123,223,227],"knowledge":[35,198],"graph":[36,46,68,87,199],"construction":[37],"bias":[39,162,203],"propagation":[40,215],"path":[41,216],"detection":[42],"method":[43,175],"based":[44,71],"on":[45,72],"neural":[47],"networks.":[48],"Four":[49],"types":[50,74],"nodes-legal":[53],"clauses,":[54],"behaviors,":[56],"algorithm":[57],"modules,":[58],"responsible":[60],"parties-are":[61],"structured":[62],"encoded,":[64],"directed":[67],"is":[69,98,112,144],"constructed":[70],"four":[73],"relationships:":[78],"\\"violation\\",":[79],"\\"basis\\",":[80],"\\"trigger\\",":[81],"\\"attribution\\".":[83],"Next,":[84],"relation-aware":[86],"convolutional":[88],"network,":[89],"combined":[90],"with":[91],"edge-type-specific":[92],"weights":[93,155],"an":[95,177,187,237],"attention":[96],"mechanism,":[97],"used":[99,113],"to":[100,114,120,130,140,158,222],"achieve":[101],"multi-layer":[102],"feature":[103],"alignment":[104,171,180],"for":[105,244],"cross-modal":[106],"semantics.":[108],"Then,":[109],"gradient-based":[110],"backpropagation":[111],"identify":[115],"edges":[116],"that":[117],"significantly":[118],"contribute":[119],"bias,":[121],"construct":[122],"bias-propagation":[124],"subgraph,":[125],"use":[127],"community":[128],"discovery":[129],"locate":[131],"root-node":[132],"clusters.":[133],"Finally,":[134],"real-time":[135],"decision":[136],"flows":[137],"are":[138,156],"mapped":[139],"temporary":[141],"nodes,":[142],"verified":[145],"embedding":[147],"similarity,":[148],"dynamics":[151],"key":[153],"edge":[154],"monitored":[157],"warn":[159],"systemic":[161],"accumulation.":[163],"Experiments":[164],"show":[165],"that,":[166],"terms":[168],"accuracy,":[172],"proposed":[174],"achieves":[176,226],"average":[178,188],"entity":[179],"accuracy":[181],"at":[183,193],"least":[184,194],"0.92":[185],"relation":[189],"preservation":[190],"completeness":[191],"0.88":[195],"across":[196],"different":[197],"density":[200],"gradients.":[201],"Regarding":[202],"localization":[204],"capability,":[205],"root":[207],"node":[208],"recall":[209],"rate":[210],"reaches":[211,218],"0.95±±0.02,":[212],"precision":[217],"0.93±±0.03.":[219],"This":[220],"research,":[221],"certain":[224],"extent,":[225],"deep":[228],"coupling":[229],"between":[230],"logic":[232],"behavior,":[235],"providing":[236],"explainable":[238],"verifiable":[240],"support":[242],"system":[243],"trustworthy":[246],"deployment":[247],"AI.":[250]}

Authors

Institutions

Publication Details

Journal
Journal of Visualized Experiments
Published
2026-09-15
DOI
https://doi.org/10.3791/72648
Primary Topic
Advanced Graph Neural Networks
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
article

Constructing Legal Knowledge Graphs and Tracing Bias in Medical Artificial Intelligence Based on Graph Neural Networks

Jun Yu
Journal of Visualized Experiments
Advanced Graph Neural Networks
article

Constructing Legal Knowledge Graphs and Tracing Bias in Medical Artificial Intelligence Based on Graph Neural Networks

Jun Yu
article en

Abstract

In medical artificial intelligence (AI) legal compliance analysis, this paper addresses the ambiguity in liability determination caused by the semantic fragmentation of multi-source heterogeneous legal texts and difficulties in tracing algorithmic biases. It constructs a knowledge graph construction and bias propagation path detection method based on graph neural networks. Four types of heterogeneous nodes-legal clauses, medical behaviors, algorithm modules, and responsible parties-are structured and encoded, and a directed graph is constructed based on four types of legal semantic relationships: "violation", "basis", "trigger", and "attribution". Next, a relation-aware graph convolutional network, combined with edge-type-specific weights and an attention mechanism, is used to achieve multi-layer feature alignment for cross-modal legal semantics. Then, gradient-based backpropagation is used to identify edges that significantly contribute to bias, construct a bias-propagation subgraph, and use community discovery to locate root-node clusters. Finally, real-time decision flows are mapped to temporary nodes, compliance is verified by embedding similarity, and the dynamics of key edge weights are monitored to warn of systemic bias accumulation. Experiments show that, in terms of semantic alignment accuracy, the proposed method achieves an average entity alignment accuracy of at least 0.92 and an average relation preservation completeness of at least 0.88 across different knowledge graph density gradients. Regarding bias localization capability, the root node recall rate reaches 0.95±±0.02, and the propagation path precision reaches 0.93±±0.03. This research, to a certain extent, achieves a deep coupling between legal logic and algorithmic behavior, providing an explainable and verifiable compliance support system for the trustworthy deployment of medical AI.

Journal of Visualized Experiments(235)
Civil Aviation Flight University of China (CN)
Peace, Justice and strong institutions
Openalex Percentile: Top 9%
Advanced Graph Neural Networks
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.