Patient-trial retrieval is a challenging problem that requires nuanced clinical reasoning beyond surface-level semantic similarity.However, scarce and costly relevance annotations force existing approaches to rely on very limited supervision or zero-shot transfer, reducing the task to generic semantic matching and failing to capture multi-factor eligibility reasoning.To this end, we propose FACTRIAL, a factorized contrastive training framework that leverages LLMs to synthesize diagnosisaware supervision for scalable patient-trial retrieval.FACTRIAL decomposes each patient note into a primary diagnosis and a set of concomitant, eligibility-triggering conditions, and constructs complementary contrastive signals through structured trial augmentation.Specifically, we generate primary-target and concomitant-target positives, together with clinically confusable near-miss negatives, to enforce diagnostic specificity under contrastive learning.Two specialized bi-encoder experts are trained to balance primary-diagnosis prioritization and concomitant-driven recall, and fused into a single deployable retriever.Experiments on three public benchmarks demonstrate that FACTRIAL achieves state-of-the-art performance, improving both top-ranked quality and high-recall coverage.

LoRa networks have been deployed in many orchards for environmental monitoring and crop management. An accurate propagation model is essential for efficiently deploying a LoRa network in orchards, e.g., determining gateway coverage and sensor placement. Although some propagation models have been studied for LoRa networks, they are not suitable for orchard environments, because they do not consider the shadowing effect on wireless propagation caused by the ground and tree canopies. This paper presents FLog, a propagation model for LoRa signals in orchard environments. FLog leverages a unique feature of orchards, i.e., all trees have similar shapes and are planted regularly in space. We develop a 3D model of the orchards. Once we have the location of a sensor and a gateway, we know the mediums that the wireless signal traverse. Based on this knowledge, we generate the First Fresnel Zone (FFZ) between the sender and the receiver. The intrinsic path loss exponents (PLE) of all mediums can be combined into a classic Log-Normal Shadowing model in the FFZ. Extensive experiments in almond orchards show that FLog reduces the link quality estimation error by 42.7% and improves gateway coverage estimation accuracy by 70.3%, compared with a widely-used propagation model.