About

Hello! I'm Guilherme Amorim, a medical doctor and clinical data scientist from Portugal. I am currently working as a Population Health Data Specialist within the Evidence Generation department at AstraZeneca.

I have over 6 years' experience working across clinical care, randomised controlled trials, and epidemiology in the UK and Portugal. I am passionate about using data to answer important questions that contribute to the common good.

You can find below a summary of my CV and a portfolio showcasing some of my work in computer science. My complete CV is

For a complete list of scientific publications please click here

Please note that all opinions and thoughts expressed in my portfolio are my own, and do not represent the views of AstraZeneca or the University of Oxford.

Thanks for visiting!

Work Experience

AstraZeneca, Lisbon (Portugal) January 2025 - present

Population Health Data Specialist (Evidence Generation)
Real-world evidenceEpidemiologyData scienceProject managementScience communicationMedical statisticsResearch methodologyClinical databases

Clinical Trial Service Unit - Oxford Population Health, University of Oxford (UK)January 2019 - January 2025

Clinical Research Fellow

Main research projects:
1. A Study of Cardiovascular Events in Diabetes (ASCEND)-PLUS (oral semaglutide for primary cardiovascular prevention in diabetes; ISRCTN76193287), 2024
2. Randomised Evaluation of COVID-19 Therapy (RECOVERY; NCT04381936), 2020-2024
3. Active Monitoring for AtriaL FIbrillation (AMALFI; remote screening for silent atrial fibrillation in primary care; ISRCTN15544176), 2019-2024
4. A Randomized Trial Assessing the Effects of Inclisiran on Clinical Outcomes Among People With Cardiovascular Disease (ORION-4; NCT03705234), 2019-2024

Main deliverables:
- Led recruitment of >5,000 participants in the AMALFI trial; contributed to recruitment of >12,000 UK participants in ORION-4 and >49,000 participants in RECOVERY
- Main author of the AMALFI statistical analysis plan
- Led NHS data linkage request submissions and linkage data collection from 27 primary care practices and centralised NHS England datasets (AMALFI and ORION-4)
- Developed data-driven algorithms to identify safety events and medication exposure (RECOVERY, ORION-4, AMALFI)
- Developed and led a centralised pre-screening procedure using linked hospital lab data (ORION-4)
- Led or contributed to 24 research papers (6 as first author), total >19,000 citations (h-index 22), including major contributions to global clinical guidelines on COVID-19 (RECOVERY trial)
- Presented research outputs at several international conferences, including European Society of Cardiology Congress 2017 and 2021

Clinical trialsEpidemiologyData scienceProject managementScience communicationMedical statisticsResearch methodologyClinical databases

Medicines and Healthcare products Regulatory Agency (MHRA), UK August 2023 - September 2023

External Medical Advisor

▪ Assessment of a trial protocol in chronic kidney disease submitted for regulatory approval .

Clinical trials Regulatory affairs Drug safety

Oxford Population Health, University of Oxford (UK) 2020 - 2024

Guest lecturer and academic supervisor, MSc in Clinical Trials

▪ Remote lectures/interviews on conducting clinical trials within primary care, medical device trials, and pilot studies.
▪ Supervised two MSc students and an Academic Foundation doctor.

Teaching Lecturing Randomised trials Clinical research

Oxford University Hospitals NHS Foundation Trust (UK) January 2019 - January 2025

Honorary/Locum Medical Doctor

▪ Admission clerking, ambulatory medical care, ward cover; total >1000 hours.

Clinical medicine Emergency care Internal medicine

Specialty Academy (Porto, Portugal)January 2018 - July 2018

Lecturer

▪ Large group lecturing for final-year medical students preparing for the national medical specialty ranking exam.

Public speaking Lecturing Gastroenterology Cardiology

Vila Nova de Gaia/Espinho Hospital Centre (Portugal)January 2018 - December 2019

General Medical Intern

▪ Rotation through multiple clinical specialties as a junior doctor, which included ward work, emergency care, and primary care clinics.

Clinical MedicineEmergency CarePrimary CareCardiology Internal Medicine

Education

Academic qualifications

Postgraduate Diploma (PgDip) in Artificial Intelligence2024 - present

Computing Department, University of Essex Online
Artificial IntelligenceData scienceMachine learningHTMLGitPythonR

Doctor of Philosophy (DPhil) in Population Health 2020 - 2024

Oxford Population Health, University of Oxford

Thesis title:
"Big drug data for big drug trials: development, validation, and implementation of routinely-collected data on medications within large trials in cardiovascular disease and COVID-19 in England and Scotland"

Complete thesis (open access)

Clinical trialsData scienceScientific writingProject managementRGitSQL

Masters in Medicine 2011 - 2017

Faculty of Medicine, University of Porto (Portugal)

Academic Performance:
Final grade 16/20 (top 11%)
93% score in the national medical specialty admission exam (top 7%)

Other courses and training

Good Clinical Practice training2024

University of Oxford

Statistical Analysis with R for Public Health Specialization2024

Imperial College London

Advanced Course in Clinical Research in Cardiology2018

Portuguese Society of Cardiology

Blended Learning Introduction to Clinical Studies2018

Faculty of Medicine of the University of Porto – Portugal

Portfolio

  • PgDip in Artificial Intelligence

    • Module 0 (Induction)

    • Module 1 (Understanding Artificial Intelligence)

    • Module 2 (Numerical Analysis)

      • Module aims: ▪ Learn about the key paradigms and algorithms in machine learning.
        ▪ Get an understanding of data analytics based on machine learning and using modern programming tools, such as Python or R.
        ▪ Experience how machine learning and data analytics can be used in real-world applications.
        ▪ Acquire the ability to gather and synthesise information from multiple sources to aid in the systematic analysis of complex problems using machine learning tools and algorithms.

        Learning outcomes: ▪ Articulate the legal, social, ethical, and professional issues faced by machine learning professionals.
        ▪ Understand the applicability and challenges associated with different datasets for the use of machine learning algorithms.
        ▪ Apply and critically appraise machine learning techniques to real-world problems, particularly where technical risk and uncertainty is involved.
        ▪ Systematically develop and implement the skills required to be effective member of a development team in a virtual professional environment, adopting real-life perspectives on team roles and organisation.

        Final grade: 79% (Distinction)

        Outputs produced:

      • Units 1-3: Collaborative Discussion 1

        Collaborative Discussion 1: The 4th Industrial Revolution

        I read the Schwab (2016) article from the World Economic Forum and discussed the impact of industry 4.0 on healthcare.


        ▪ Initial Post

        ▪ Summary Post

        Here are my contributions to other students' posts:

        ▪ Peer Response on Martyna's Initial Post

        ▪ Peer Response on Dinh Khoi Dang's Initial Post

        ▪ Peer Response on Jafar's Initial Post

        Literature searchCritical thinking

      • Unit 2: Seminar Preparation

        Exploratory data analysis

        I performed exploratory data analysis and cleaning in the Kaggle auto-mpg dataset, including the following steps:
        ▪ Identify missing values.
        ▪ Estimate Skewness and Kurtosis.
        ▪ Correlation Heat Map.
        ▪ Scatter plot for different parameters.
        ▪ Replace categorical values with numerical values (i.e., America 1, Europe 2 etc.).


        Jupyther notebook here:

        Python Exploratory data analysis Data visualisation Data cleaning

      • Unit 3: E-Portfolio Component

        Practical exploration of Python code for linear and polynomial regression.

        Exercise 1: correlation (with some changes to data points made by myself)

        Exercise 2: simple linear regression (with some changes to data points made by myself)

        Exercise 3: multiple linear regression (no changes made, just read through and explored the code provided)

        Exercise 4: polynomial regression (no changes made, just read through and explored the code provided)

        PythonCorrelationLinear regressionPolynomial regression

      • Unit 4: Seminar Preparation

        Linear regression with Scikit-Learn

        In this Unit, I first worked through the Scikit-Learn linear regression coding example provided. Then I used global data from to assess the relationship between population and gross domestic product (GDP) per capita, using correlation and linear regression (and performing exploratory data analysis and data cleaning for that end).
        Jupyter notebook

        Python Data processing Correlation Linear regression Scikit-Learn Model evaluation

      • Unit 5: Wiki Activity

        K-means clustering

        Wiki entry describing K-means clustering algorithm, as well as its streghts and pitfalls, based on experimentation with two interactive animations.

        K-means clusteringUnsupervised machine learningCritical thinking

      • Unit 5: E-Portfolio Component

        Jaccard coefficients

        I calculated Jaccard coefficients (to assess similarity/dissimirity between observations) using a practical example.

        ClusteringJaccard coefficientsPerformance evaluation

      • Unit 6: Seminar Preparation

        K-means clustering (practical application)

        I read through the example K-means clustering notebook provided, and applied this clustering technique in different settings using the 3 datasets suggested.
        Iris dataset
        Wine dataset
        Australian weather dataset


        PythonK-means clusteringUnsupervised machine learning

      • Unit 7: E-Portfolio Component

        Artificial Neural Networks (practical application)

        I worked through the exercise notebooks provided and added comments to facilitate interpretation.
        Simple perceptron
        Perceptron with AND operator
        Multilayer perceptron

        PythonArtifical Neural Networks

      • Unit 8: E-Portfolio Component

        Gradient descent (practical application)

        I worked through the exercise notebook provided, and added comments and a visusalisation of the cost function to facilitate interpretation.
        Gradient descent

        Python Artifical Neural Networks Gradient descent

      • Unit 8: Seminar Preparation

        Emerging research in Artificial Neural Networkds

        I read the articles by Pruciak (2021) and the Centre for Data Ethics and Innovation (2019), and then wrote down some thoughts on the application of Artifical Neural Networks in healthcare, and my concerns about the use of AI-based technology in the insurance industry.

        Artifical Neural NetworksCritical thinkingArtifical Intelligence in industry

      • Unit 8-10: Collaborative Discussion 2

        Risks and benefits of AI writers

        I read the article by Hutson (2021) and wrote about the benefits and dangers of large language model (LLMs) applicationsacross a range of industries and settings, as well as possible risk mitigating approaches.

        Initial post

        Summary post

        Peer response on Martyna's initial post

        Peer response on Dinh Khoi Dang's initial post

        Large language modelsCritical thinkingLiterature review

      • Unit 9: E-Portfolio Component

        Ethical and social implications of convoluted neural networks

        I read the article by Wall (2021) and wrote about my thoughts on the ethical and social implications of convoluted neural networks (CNNs).

        I also went through the Jupyter notebook provided on image recognition using a simple CNN, changed the input image and verified the prediction being made (NB the notebook is not my work, I just changed some parameters for practice purposes).

        CNNsCritical thinkingImage recognition

      • Unit 10: Seminar Preparation

        Convoluted neural network architecture

        I explored the architecture of CNN networks using the CNN Explainer visualisation and associated article, and tested how the algorithm performed with images belonging to classes it does know (e.g. orange) vs classes it doesn't (e.g. motorbike)

        CNNsReLu function Softmax function Image recognition Critical thinking

      • Unit 11: E-Portfolio Component

        Model performance evaluation

        I explored the notebook provided with practical examples of performance evaluation metrics and their calculations, wrote some annotations, and changed some data points to see their impact on prediction and performance (the underlying code is not mine).

        Model performance evaluationScikit-learn

      • Unit 12: Seminar Preparation

        Future of Machine Learning

        I read the article by Diez-Olivan et al (2019) and noted down some thoughts on use-cases and challenges of prescriptive machine learnign models within healthcare.


        Industry 4.0Critical thinking

      • Group assignment - analytical report

        Airbnb business analysis using a data science approach

        In this group assignment, we used data from the Kaggle New York City Airbnb dataset to produce a business report. Our analysis produced insights into price drivers, and a predictive model which can be used to support pricing strategies and identify over- and underperforming listings.

        Business report
        Python code and outputs
        Collaborative GitHub repository


        Grade: 75% (Distinction)

        Critical thinking Python Git Exploratory data analysis Visualisation Linear regression Performance evaluation Jupyter notebooks Agile procedures Project management Collaborative coding

      • Individual assignment - image classification presentation

        Image classification using convolutional neural networks

        In this individual assignment, I developed and trained multiple convolutional neural network models for image classification in the CIFAR-10 dataset and produced a video presentation. My development approach covered diverse architectures, hyperparameters, and training approaches. I explored how to tackle common challenges involded in developing machine learning models, and critically assessed the performance on my final model on test data.

        Slide presentation
        Audio transcript
        Python code and outputs


        Grade: 81% (Distinction)

        Critical thinking Python Keras Convolutional neural networks Computer vision

      • Assignment - reflective piece

        Critical reflection on my learning journey throughout this module

        Full report here

        Grade: 80% (Distinction)

        Critical thinking Reflective learning Machine learning Collaborative development Python Git HTML CSS Jupyter notebooks
      • Module aims: ▪ Learn about the key paradigms and algorithms in machine learning.
        ▪ An understanding of the motivations for, and appropriate use of, agent-based computing.
        ▪ An understanding of the main agent models in use today and their grounding in artificial intelligence research.
        ▪ The knowledge and skills required to develop, deploy and evaluate the tools and techniques of intelligent systems to solve real-world problems.
        An understanding of contemporary research issues in the area of intelligent agent systems.

        Learning outcomes: ▪ Identify and critically analyse agent-based systems, differentiating between architectures and approaches.
        ▪ Apply and critically evaluate intelligent agent techniques to real-world problems, particularly where technical risk and uncertainty is involved.
        ▪ Deploy critically appropriate software tools and skills for the design and implementation of an agent-based system, bearing in mind applicable legal, social, ethical and professional issues.
        ▪ Systematically develop and implement the skills required to be effective member of a development team in a virtual professional environment, adopting real-life perspectives on team roles and organisation.

        Final grade: 79% (Distinction)

        Outputs produced:

      • Units 1-3: Collaborative Discussion 1

        Collaborative Discussion 1: Agent Based Systems

        I discussed the factors leading to the rise of agent-based systems and possible benefits and risks of this approach to organisations.


        ▪ Initial Post

        ▪ Summary Post

        Here are my contributions to other students' posts:

        ▪ Peer Response on Georgios' post

        ▪ Peer Response on Yemi's post

        Agent-based computing Critical thinking AI ethics

      • Units 5-7: Collaborative Discussion 2

        Collaborative Discussion 2: Agent Communication Languages

        I discussed the pros and cons of using specific agent communication languages (such as KQML) vs traditional method invocation in standard languages (such as Python or Java) for communication among agents in multi-agent systems.


        ▪ Initial Post

        ▪ Summary Post

        Here are my contributions to other students' posts:

        ▪ Peer Response on Nooras' post

        ▪ Peer Response on Haris' post

        Agent-based computing Agent communication languages KQML Python Multi-agent systems

      • Unit 6: E-Portfolio Component

        Creating agent dialogues
        I used KQML and KIF to create a simple dialogue between two agents.

        Agent communication languagesKQMLKIF

      • Unit 8: E-Portfolio Component

        Creating Parsing Trees

        In this unit, I created constituency-based parsing trees to parse three different dialogues using natural language processing.

        Natural language processing Parsing trees

      • Units 9-11: Collaborative Discussion 3

        Collaborative Discussion 3: Deep learning

        I discussed the potencial opportunities offered by emerging trends in deep learning models, especially generative AI content, but also the associated risks and ethical considerations.


        ▪ Initial Post

        ▪ Summary Post

        Here are my contributions to other students' posts:

        ▪ Peer Response on Noora' post

        ▪ Peer Response on Haris' post

        Deep learning Generative AI Critical thinking AI ethics

      • Unit 10: E-Portfolio Component

        Deep learning in action

        I explored the potencial use of deep learning algorithms in healthcare, especially for predictive medicine approaches, and discussed how the technology can be applied, its possible benefits to society, but also the underlying risks and ethical issues.

        Deep learningBig data Predictive modelling AI ethics

      • Group assignment - development project report

        Online academic research agent

        In this group assignment, we produced a concept development report for an intelligent multi-agent academic research assistent, capable of finding results on a website based on search terms (e.g., social media or a search engine), extracting the data, and sending to an offline location.
        Development report


        Grade: 86% (Distinction)

        Agent-based computing Multi-agent systems Python Langchain Data schemes Project management Collaborative development

      • Individual assignment - agent-based systems

        Complete Python implementation

        In this assignmment, I proceeded to produce a coded implementation of the concept articulated in the team development report, using Python. I was able to develop, test, and showcase a fully-functioning software tool composed of multiple agents working in sequence, and using a circular control flow system. I learnt how to employ Langchain for streamlined and scalable LLM-based systems, developed a simple graphical user interface, and implemented a structured SQL database with additional vector indexing features. Finally, I critically assessed my implementation and identified some strenghts and weaknesses, as well as areas for potencial improvement.

        Slide presentation
        Audio transcript
        Jupyter notebook (development and testing)


        Grade: 82% (Distinction)

        Python Langchain OpenAI models GUI SQLite3 Vector indexing Prompt engineering Testing Multi-agent systems Critical thinking

      • Assignment - reflective piece

        Critical reflection on my learning journey throughout this module

        Full report here:

        Grade: 72% (Distinction)

        Critical thinking Reflective learning Agent-based systems Collaborative development Team management Python
      • Module aims: ▪ Discuss how to define knowledge and explain the concept of KRR and its relationship to data and information.
        ▪ Gain an appreciation of the importance and rationale for employing formal approaches to KRR.
        ▪ An understanding of the main agent models in use today and their grounding in artificial intelligence research.
        ▪ Demonstrate an understanding of the basics of logic and its application to KRR problems.
        ▪ Acquire an understanding of the professional competence required to able be to identify and examine KRR technologies for Artificial Intelligence applications.

        Learning outcomes: ▪ Critique the need for formal approaches to knowledge representation and reasoning.
        ▪ Review critically properties of a knowledge-based system.
        ▪ Appraise critically modelling techniques for knowledge representation and reasoning.
        ▪ Examine and incorporate different modelling approaches to solving KRR problems.


        Outputs produced:

      • Units 1-3: Collaborative Discussion 1

        Collaborative Discussion 1: Knowledge Representation and Reasoning - historical and semantic considerations

        I discussed whether knowledge representation could be considered a novel or ancient concept, and how it relates to knowledge reasoning

        ▪ Initial Post

        ▪ Summary Post

        Here are my contributions to other students' posts:

        ▪ Peer Response on Nikolaos' post

        ▪ Peer Response on Abdulrahman's post

        ▪ Peer Response on Georgios' post

        Knowledge representation Knowledge reasoning History

      • Unit 1: Formative Activity

        Reflection on knowing vs having information about

        Knowledge representation and reasoning

      • Unit 2: Seminar Preparation

        Practical exercise: sets, set Theory, truth Tables and logic


        Logic inference Set theory Truth tables

      • Unit 3: Formative Activity

        Practical exercise: first-order logic

        First-order logic

      • Unit 4: Seminar Preparation

        Practical exericse: introduction to logic programming using Prolog

        First-order logic Prolog

      • Unit 5: Formative Activity

        Reflection on the definition of knowledge bases and knowledge-based economies

        Knowledge-base Knowledge-based economies

      • Unit 6: Formative Activity

        Reflection on ontology development approaches

        Ontologies Ontology development

      • Unit 8: Formative Activity

        Practical exercise: ontology creation using Protégé

        Ontologies Ontology development Protégé RDF OWL

      • Unit 10: Formative Activity

        Practical exercise: inference and knowledge modelling using Protégé

        Ontologies Ontology development Protégé RDF OWL

      • Unit 11: Seminar Preparation

        Reflection on ontology design principles and evaluation

        Ontologies Ontology development Ontology evaluation

      • Units 9-11: Collaborative Discussion 2

        Collaborative Discussion 2: Ontology definitions and computer languages

        I analysed a possible definition of an ontology, and discussed the pros and cons of using different computer languages (such as OWL2) fpr building ontologies and modeling knowledge bases

        ▪ Initial Post

        ▪ Summary Post

        Here are my contributions to other students' posts:

        ▪ Peer Response on Georgios' post

        ▪ Peer Response on Nikolaos' post

        Ontologies Ontology-modelling languages KIF OWL2 OWL-lite RDF

      • Case Study Review

        Analysis and Application of the Intelligence Task Ontology (ITO) in AI Benchmarking

        Grade: 72% (Distinction)

        Ontology development Ontology evaluation Real-world applications Critical analysis

      • Individual assignment - knowledge modelling

        Complete Python implementation

        In this assignmment, I conceptualised, developed, implemented, and evaluated an ontology aimed at supporting an AI-driven job matching algorithm, capable of linking suitable candidates and jobs based on industry, experience, qualifications, and preferred location and salary.

        Protégé Ontology development Ontology evaluation

      • Assignment - reflective piece

        Critical reflection on my learning journey throughout this module

        Full report here:
        Critical thinking Reflective learning

Contact

You can contact me at guilhermepessoaamorim@gmail.com