让医学语言模型别把无关变量当因果，靠人属性嵌入纠正错误关联。
Correlation Is Not Enough: Embedding Human Metadata for Individual Causal Discovery
- 用人类行为数据增强模型嵌入，避免跨域无关项误判为相关
- 经两阶段优化后，跨域区分度提升至2.3倍，准确率提高0.392
- 适用于医疗因果推断、个体化健康建模等需严谨关联分析的场景
询问预训练生物医学语言模型‘皮质醇28 ug/dL’与‘股市波动’是否相关，其余弦相似度达0.83，实则无机制关联。我们测试的多个生物医学编码器（BioBERT、PubMedBERT、BioM-ELECTRA）对跨域无关对的得分均在0.76至0.92之间，而真实答案应接近零，跨域判别准确率为0%。检索系统可幸免，因下游有噪声过滤；但以人为对象的大行为模型（LBM）无法回避：它将嵌入邻近性视为因果证据，错误邻近导致虚假因果边传播。本文提出修复方案：首先对72,034对样本进行对比学习，使PubMedBERT的BIOSSES相关性从0.633升至0.828，跨/内域分离度从1.05x增至1.63x；第二步BODHI从生物医学知识图谱中挖掘缺失边作为难负例，进一步将分离度提升至2.30x，判别差距+0.392，仅增加4.5%的BIOSSES成本。在Intel Xeon 6737P（AMX）上，OpenVINO将单次查询延迟从1367毫秒降至10毫秒（133倍加速），吞吐率达555句/秒。意外发现：此硬件下FP16优于INT8，且无论批量大小均胜出。无AMX的Ice Lake实例慢13-27倍。代码、基准、语料库、BODHI生成器及OpenVINO脚本均已开源。
原文摘要 · Abstract (English)
Ask a pretrained biomedical language model whether "cortisol 28 ug/dL" and "stock-market volatility" are related, and it returns a cosine similarity of 0.83 on a scale where 1.0 means identical. The two share no mechanism. This is not a corner case: every off-the-shelf biomedical encoder we tested (BioBERT, PubMedBERT, BioM-ELECTRA) scores unrelated cross-domain pairs between 0.76 and 0.92 when the answer should be near zero. Accuracy on cross-domain discrimination is 0%. Retrieval systems survive this, because a language model downstream filters the noise. A Large Behavioural Model (LBM), a foundation model whose subject is a person rather than a sentence, does not: it reasons over a graph of a user's life and treats embedding proximity as evidence that two events are causally linked. False proximity writes a false causal edge, and everything downstream inherits the error. Here, embedding geometry is not a tuning knob; it is correctness. We report the fix. A contrastive pass over 72,034 pairs raises PubMedBERT BIOSSES correlation from 0.633 to 0.828 and within-vs-across-domain separation from 1.05x to 1.63x. A second pass, BODHI, mines hard negatives from edges absent in a biomedical knowledge graph and lifts separation to 2.30x and the discrimination gap to +0.392, at a 4.5% BIOSSES cost. On an Intel Xeon 6737P with AMX, OpenVINO cuts single-query latency from 1367 ms to 10 ms (133x) and reaches 555 sentences/sec. One finding contradicts standard advice: FP16 beats INT8 on this silicon at every serving batch size, and we explain why. The same model on a no-AMX Ice Lake instance runs 13-27x slower. We release the benchmark suite, training corpora, the BODHI generator, and the OpenVINO scripts.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。