Ryan T. Hodgson

Ryan T. Hodgson


Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications

14 May 2013 in Internal Medicine Journal
Verified
Journal Article
Open access
PDF file
RHRyan T. Hodgson
Ryan T. Hodgson
Click on the article title to read more.

9 July 2021
Verified
Book Chapter
Closed access
PDF file
RHRyan T. Hodgson
Ryan T. Hodgson
Alexandra I. CristeaAlexandra I. Cristea
Alexandra I. Cristea
LSLei Shi
Lei Shi
JGJohn Graham
John Graham

22 May 2024 in IEEE Access
Verified
Journal Article
Open access
RHRyan T. Hodgson
Ryan T. Hodgson
JWJingyun Wang
Jingyun Wang
Alexandra I. CristeaAlexandra I. Cristea
Alexandra I. Cristea
JGJohn Graham
John Graham
Real-world applications of word embeddings to downstream clustering tasks may experience limitations to performance, due to the high degree of dimensionality of the embeddings. In particular, clustering algorithms do not scale well when applied to highly dimensional data. One method to address this is through the use of dimensionality reduction algorithms (DRA). Current state of the art algorithms for dimensionality reduction (DR) have been demonstrated to contribute to improvements in clustering accuracy and performance. However, the impact that a neural network architecture can have on the current state of the art Parametric Uniform Manifold Approximation and Projection (UMAP) algorithm is yet unexplored. This work investigates, for the first time, the effects of using attention mechanisms in neural networks for Parametric UMAP, through the application of network architectures that have had considerable effect upon the wider machine learning and natural language processing (NLP) fields - namely, the transformer-encoder, and the bidirectional recurrent neural network. We implement these architectures within a semi-supervised metric learning pipeline, with results demonstrating an improvement in the clustering accuracy, compared to conventional DRA techniques, on three out of four datasets, and comparable SoA accuracy on the fourth. To further support our analysis, we also investigate the effects of the transformer-encoder metric-learning pipeline upon the individual class accuracy of downstream clustering, for highly imbalanced datasets. Our analyses indicate that the proposed pipeline with transformer-encoder for parametric UMAP confers a significantly measurable benefit to the accuracy of underrepresented classes.

18 February 2026 in BMC Medical Education
Verified
Journal Article
Open access
PDF file
JWJingyun Wang
Jingyun Wang
RHRyan T. Hodgson
Ryan T. Hodgson
MLMarie Shigematsu Locatelli
Marie Shigematsu Locatelli
Background: For medical professionals, the comprehension and analysis of literature is essential, however, it requires time which could be dedicated to the primary duties of providing care to patients. To address these conflicting needs and the challenge of finding new and appropriate information at the right time during medical literature review analysis, we introduce, to the best of our knowledge, the first automated framework co-designed with medical professionals , utilising a human-AI hybrid approach for the consolidation of large volumes of literature into manageable, topic-based chunks. Methods: For this framework (named SMARTEN), on the AI side, cluster-based topic modelling is deployed for the analysis of academic articles (consisting of abstracts and full article content) extracted from the PubMed database, the go-to venue for medical professionals, to ensure the provision of a wide range of topics reflecting this rich field and access to large volumes of biomedical literature. After consolidating topics within the literature, semantic analyses of the relationships between topics are performed, based on the content of the literature within each topic, ultimately assisting in the discovery of new knowledge from the literature. The effectiveness of the proposed approach was evaluated through a software implementation of the SMARTEN framework ( the SMARTEN system ), and a subsequent experimental study upon the concepts of cognitive load, technology acceptance, and general satisfaction by 18 medical practitioners and medical students. Moreover, the labels assigned by participants to publications were collected and analysed to explore the average accuracy for identifying relevant publications in the SMARTEN system. Results: Overall, our analysis of user feedback and user logs indicates that the SMARTEN systsem contributed to enhancing the discovery of new knowledge within the literature, while maintaining cognitive load at manageable levels for medical professionals and students. Furthermore, our SMARTEN system effectively identifies relevant publications that would typically appear far down in PubMed search rankings, consistently positioning them within the top 10 results. Conclusion: This study presents the SMARTEN framework for the automated analysis of large volumes of academic literature based on the semantic contents of individual publications via cluster-based topic modelling. This enables the expediting of literature review for busy medical experts, while also enabling the discovery of new knowledge which would otherwise have gone unnoticed in the wealth of research available online. We evaluate the system developed based on this framework in an experimental study with medical experts, to reinforce the benefit that this framework provides and to identify improvements for further research.