Dr Samantha Pendleton
Engineer of data, ontologies, and clusters.
Thrower of pots, controllers, and eggs.
What's the Catch with ChatGPT?
Validation of NLP tools and comparison to LLMs ChatGPT. Now that I have your attention. This isn’t about ChatGPT specifically but rather about the larger picture: Large Language Models (LLMs). ChatGPT is a large language model.
Introduction Many Natural Language Processing (NLP) tools are perhaps becoming obsolete with the emergence of complex LLMs, most notably ChatGPT.
LLMs have been around for a while but gained attention around 2020 when GPT3 was released by OpenAI1. LLMs are designed to understand human language and are trained on large amounts of text: books, code, social media, clinical documentation, transcripts, and more. These models learn the patterns in language in order to “communicate”. When asked a question, the model observes the tokens (words) you used and based on its knowledge, it’ll generate a response by predicting/arranging tokens in such a way that resembles a conversation.
AI-generated depiction of my PhD
AI-generated depiction of my PhD Using the DALL·E mini model, I threw my PhD at it and this is what it threw back. Looks sort of like cells or a flexible ontology. Check out the model on Huggingface.
The PhD experience
Becoming a Doctor in Clinical Informatics The information in this post is my personal PhD journey. I aim to inform and share my PhD (experience and research) with you – the reader. Year 1 (Oct 2018-) You don’t have as much experience as I would have thought…
This was one of the first things said to me. By a postdoc. Imposter syndrome was planted and settling in for the long haul.
Hello World
Version 3 has been released I started this website journey around 2015 for university assignments: writing blogs for modules. I decided to create my own CSS and index page to better structure my work. In 2021, midway through my PhD, I progressed onto GitHub pages with Bootstrap and a terrible repo structure. Now in 2023, I’ve decided to try Hugo.
A few months ago, with COVID, I decided to revamp the site and intended to go with the markdown-file route. Whilst ill with the decision to improve the blog, I had some time off work, so my 5-day isolation weekend turned somewhat productive.
Investigations into the patient voice: a multi-perspective analysis of inflammation
Investigations into the patient voice: a multi-perspective analysis of inflammation This work is my PhD1, completed in 2023.
Abstract Background The patient is the expert of their medical journey and their experiences go largely unheard in clinical practice. Understanding the patient is important as bridging gaps in the medical domain enhances clinical knowledge: benefiting patient care in addition to improving quality of life.
My PhD presented challenges and solutions exploring quality of life pertaining to two inflammatory diseases, Uveitis and Inflammatory Bowel Disease (IBD), which are often undifferentiated. My work explored these inflammatory conditions and the patient’s voice through various methods, including sentiment analysis, semantic characterisations, and clustering.
Fashun
Fashun: an R Shiny App of fashion trends I want to start by saying fashion (or fashun) is whatever you decide - whether that is double denim or choosing two completely different patterns. I enjoy fashion, and for me, this is an outlet for expressing oneself.
» Fashun App
Introduction Setting the scene: I have just handed in my PhD thesis and although I have a job, my evenings are no longer full of writing. In my job, I have been developing a data dashboard and found I’ve enjoyed using Shiny. So, I decided I would expand my skills in these evenings: looking at data handling, dashboards, and text analysis.
The Patient Voice as a Clinical Narrative via Embeddings: PatientINF & COID
PatientINF & COID Resources in this project are both are available open-source for secondary research use:
PatientINF is an embedding model. COID is the Combined Ontology for Inflammatory Diseases. Background Semantic similarity builds upon the patient voice in understanding synonyms and the sort of terms patients are using. Not only can semantics help to better undertand the patient, but also the clinician. Patients may not use clinical terms (jargon) often and domain experts may be unfamiliar with patient-preferred terms.
Ocular Immune-Mediated Inflammatory Diseases ontology (OcIMIDo)
Development and application of the Ocular Immune-Mediated Inflammatory Diseases ontology (OcIMIDo) enhanced with synonyms from online patient support forum conversation OcIMIDo is the first ontology of its kind in opthamology1. Available open-source on GitHub and other biomedical ontology platforms, including Bioportal.
Background Uveitis and other ocular immune-mediated inflammatory diseases (IMIDs) are chronic conditions (usually life-long) and in many cases, life threatening. Uveitis itself is one of the top causes of blindness and often undifferentiated: no identifiable cause.
Jabberwocky: an NLP toolkit for those nonsensical ontologies
Jabberwocky Jabberwocky is a Natural Language Processing (NLP) toolkit for those nonsensical ontologies1. Available open-source on GitHub.
Unstructured text is a valuable resource for research, yet text mining is a complicated task. Text mining with key terms can be limited and potentially enhanced with the knowledge of synonyms. Especially synonyms in that domain.
Ontologies are useful - they condense a domain of knowledge in a structured manner and classes may have synonyms. Ontologies have proven useful in text mining tasks, yet there lie gaps in the NLP community for the easy manipulation of ontologies.
ResearcHers Code takeover
I ran the ResearcHers Twitter account from the 8th until the 12th February 2021 - below are Tweets I copied over and edited/shortened for the blog/reader. Monday: history I’m a 3rd year PhD student (2021) at the University of Birmingham, UK. My field is Clinical Informatics, and I am looking at inflammation! Today I’ll talk about my experience learning to code throughout education: when it started, how it progressed, and what I can do now!