Biology subjects

Soar, P.

Publications and source records attributed to Soar, P..

3 recordsLinked to original sources

bioRxiv2026

BoneGraph: A Domain-Specialised, Self-Correcting Reasoning System for Bone Science Retrieval, Grounded Inference, and Image Mechanics

Bone science literature spans biology, mechanics, materials science, and clinical medicine, and its volume makes reliable knowledge synthesis increasingly difficult. General-purpose large language models (LLMs) answer fluently but under-represent this niche domain, cannot cite specific evidence, and offer no mechanism to be corrected durably. Here we present BoneGraph, a domain-specialised system for bone science delivered as a five-tab web application over a shared substrate: a curated full-text corpus of 7,449 documents embedded into 248,629 passage vectors using SPECTER2, a scientific-paper embedding model, and a bone knowledge graph of 1,597 concepts with 1,699 causal relations. The five tabs are: (I) Chat, retrieval-augmented question answering with server-rebuilt inline citations; (II) Search, raw semantic retrieval with no LLM in the loop; (III) Reasoning, a self-correcting loop in which a deterministic physics check and a literature/knowledge-graph critic constrain the answer, and a user's feedback becomes a durable, per-user rule; (IV) Vision, a bone-region classifier trained on frozen BiomedCLIP features that grounds a vision-language model, guarded against out-of-distribution inputs and augmented with image-embedding correction memory; and (V) Mechanics, integrating our previous data-driven image mechanics (D2IM) model that predicts displacement and strain fields from a single undeformed micro-CT image. All inference is performed locally, without third-party API calls, and the public beta is served at bonegraph.org. Retrieval attains a mean reciprocal rank (MRR) of 0.928 on a 30-question, seven-domain benchmark, and the Vision classifier attains 92.6% accuracy on the held-out MURA (MUsculoskeletal RAdiographs) dataset. A grounded-reasoning benchmark shows that, with the correct passage, BoneGraph raises answer accuracy from 42% to 78%. BoneGraph makes a major contribution to bone-science informatics: to our knowledge it is the first domain-specialised system to unify curated retrieval, deterministic physics-grounded self-correction, and durable per-user learning for bone science.

bioengineering↗

bioRxiv2026

Evaluation of direct strain field prediction in bone with data-driven image mechanics (D2IM-Strain)

Digital volume correlation (DVC) has become the benchmark experimental technique for full-field strain measurement in bone mechanics. In our previous work we developed a novel data-driven image mechanics (D2IM) approach that learns from DVC data and predicts displacement fields directly from undeformed X-ray computed tomography (XCT) images, deriving strain fields from such predictions. However, strain fields derived through numerical differentiation of displacement fields amplify high-frequency noise, and regularization techniques compromise spatial resolution while incurring substantial computational costs. Here we propose the upgrade D2IM-Strain to predict strain fields directly from XCT images of bone. Two prediction strategies were compared: displacement-derived strain and direct strain prediction. The direct strain prediction model significantly improved accuracy particularly for strain magnitudes below 10000{varepsilon}, taken as a representative threshold value for bone tissue yielding in compression. In addition, the direct approach reduced false-positive high-strain classifications by 75%. By eliminating numerical differentiation, the approach reduces noise amplification while maintaining computational efficiency. These findings represent a critical step toward developing robust data-driven volume correlation methods for hierarchical materials.

bioengineering↗

bioRxiv2023

Data-driven image mechanics (D2IM): a deep learning approach to predict displacement and strain fields from undeformed X-ray tomography images - Evaluation of bone mechanics.

Experimental measurement of displacement and strain fields using techniques such as digital volume correlation (DVC) from in situ X-ray computed tomography (XCT) has notably advanced the understanding of bone mechanics from organ to tissue level. Being experimental in nature, DVC output has been often employed to validate finite element (FE) models of bone improving their predictive ability. Despite the excellent results achieved, these techniques are complex, time consuming, potentially affecting tissue mechanical properties, and their predictive ability requiring prior knowledge of material properties. The recent advent of deep learning (DL) has enabled data-driven models, paving the way for the full exploitation of rich image datasets from which physics can be learnt and retained. Here we propose a novel data-driven image mechanics (D2IM) approach based on feed forward convolutional neural network (CNN) that learns from DVC displacement fields of vertebrae, predicting displacement and strain fields for undeformed XCT images. D2IM successfully predicted all displacement fields, particularly the one for the z loading axis (w), where high correlation (R2=0.93) and minimal error (as low as less than 1m) were found when comparing measured against predicted displacements. The predicted axial strain field in z ({varepsilon}zz) was also consistent in distribution with the measured one, displaying generally reduced errors (as low as few tens of {varepsilon}) in the regions within the vertebral body where the effect of border outliers was minimal. This is the first study using experimental full-field measurements on bone structures from DVC to inform DL-based model such as D2IM, which represents a major contribution in the prediction of displacement and strain fields only based on the greyscale content of undeformed XCT images. The future development of D2IM will incorporate a wider range of structures and loading scenarios for accurate prediction of physical fields in both hard and soft tissues, aiming at clinical translation for improved diagnostics.

bioengineering↗