This repository contains prototype systems exploring how artificial intelligence could support clinical trial operations, regulatory workflows, and healthcare technology systems.
The projects focus on designing AI-assisted tools for regulated environments, where transparency, traceability, and human oversight are essential.
These prototypes demonstrate how domain expertise from clinical research can be combined with modern AI techniques to design responsible health technology systems.
Clinical trial protocols are lengthy documents that must be reviewed by clinical operations, regulatory teams, and investigators.
This prototype demonstrates how large language models can convert unstructured protocol text into structured trial metadata.
Key capabilities:
- structured extraction of trial metadata
- eligibility criteria identification
- endpoint extraction
- confidence scoring
- evidence-linked outputs
- human validation flags
This workflow illustrates how AI could assist protocol review while maintaining regulatory oversight.
Drug safety monitoring requires analysing large volumes of adverse event reports to detect potential safety signals.
This project demonstrates how AI-assisted pattern detection could identify clusters of similar adverse events requiring expert investigation.
Concepts explored:
- adverse event clustering
- safety signal detection
- pharmacovigilance workflows
- human safety review
Selecting investigator sites is one of the most important decisions in clinical trial operations.
This prototype demonstrates how machine learning could predict site recruitment performance using historical operational data.
Key components:
- feature engineering from site performance data
- machine learning prediction models
- explainable feature importance
- operational risk drivers
- human validation of model outputs
Informed consent documents can be complex and difficult for patients to understand.
This prototype explores how AI could improve participant comprehension by generating plain-language explanations and comprehension checks.
Concepts demonstrated:
- consent document summarization
- structured consent extraction
- teach-back comprehension questions
- human review workflows
- governance-aware AI design
These projects explore several key principles for AI systems operating in healthcare environments:
Human-in-the-loop AI
AI outputs should support experts rather than replace them.
Evidence traceability
Extracted information should link directly to source text.
Structured data workflows
AI outputs should be machine-readable and auditable.
Governance and oversight
AI systems must incorporate regulatory awareness and validation layers.
- Large Language Models (LLMs)
- Prompt engineering
- Structured data extraction
- Machine learning prediction models
- AI system architecture design
- Human-in-the-loop AI workflows
- Health technology product thinking
My background is in clinical research operations and regulatory processes.
Through this work I became increasingly interested in how AI systems could improve healthcare operations, particularly in areas such as:
- clinical trial workflows
- safety monitoring
- regulatory documentation
- patient-facing health technologies
This portfolio explores how those systems could be designed responsibly.
These projects are conceptual demonstrations using public or simulated example data.
They are intended to illustrate AI system design concepts and should not be used for clinical or regulatory decision-making.