THBKG: A Temporal Biomedical Knowledge Graph for Decision-Aligned Clinical Advancement Prediction

2026-08-06

AI Summary

Current biomedical knowledge graphs lack a temporal dimension, limiting their ability to predict whether target–disease associations can advance to Phase III clinical trials based on historical evidence. This work proposes a time-aware heterogeneous knowledge graph, THBKG, which uniquely reconstructs the full evidence landscape as of specific decision time points. By leveraging graph neural networks, the framework propagates biological signals to make predictions even in the absence of direct evidence. Integrating a temporally aligned benchmark design with path-based interpretability analysis, the method achieves a 4.3–4.5-fold increase in success rate among the top-10 predictions across therapeutic areas. Notably, for 72.8% of target–disease pairs lacking direct evidence, the model’s performance surpasses random chance by 5–6 times.

Abstract

Inadequate target--disease linkage accounts for 40--50\% of Phase~II efficacy failures, so anticipating which programmes will advance would let sponsors back the hypotheses most likely to reach patients. What a programme can be judged on is the evidence that supported its linkage \emph{when it entered the clinic}. No existing biomedical knowledge graph allows that evidence profile to be assembled as of a past date. We present the Temporal Heterogeneous Biomedical Knowledge Graph (THBKG), which describes and predicts therapeutic target--disease links through time: 110,396 entities and 11.1M edges across nineteen relation types, each edge carrying the year its evidence changed, so a pair's profile can be recovered as it stood when its own decision fell due. On this graph we define a decision-aligned benchmark that predicts, for a target--disease pair entering Phase~II, whether it advances to Phase~III on evidence datable before that decision. Graph propagation over the THBKG outranks every direct-evidence reference scored under the same decision-aligned protocol, reaching a relative success of 4.3--4.5 at the top ten pairs per therapeutic area. The gain concentrates on the 72.8\% of pairs with no direct target--disease evidence at their decision point, where a direct-edge model has nothing to read: the encoders still rank five- to sixfold above chance, recovering the signal by propagating over the intervening biology. Adapting a path-based explainer to the decision-time subgraph decomposes each prediction into the evidence landscape behind the hypothesis for explainable prediction. We release the THBKG as a continually updated substrate for studying therapeutic target hypotheses by retrospective validation.

Problem

• temporal knowledge graph
• target-disease linkage
• clinical advancement prediction
• evidence profiling
• decision-aligned benchmark

Innovation

• Temporal Knowledge Graph
• Decision-Aligned Prediction
• Graph Propagation
• Therapeutic Target Validation
• Explainable AI