WAG builds a query-adaptive knowledge graph from wearable data using hierarchical Bayesian modeling to retrieve relevant context for LLM reasoning and reports ~70% win rate over baselines.
"" 2Generate clinically relevant questions from wearable data, with each question containing 2-3 metrics. 3 4INPUT FORMAT (Array of metric objects): 5[ 6{ 7
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Query-Conditioned Graph Retrieval for Contextualized LLM Reasoning in Personalized Wearable Data
WAG builds a query-adaptive knowledge graph from wearable data using hierarchical Bayesian modeling to retrieve relevant context for LLM reasoning and reports ~70% win rate over baselines.