Joshua Yeats
Simulated CGM — Keep it in range
Selected Work
- 01
Honours DissertationLearning from Sparse Data
Undergraduate thesis evaluating program synthesis, LLM rule generation, and reinforcement learning for interpretable decision support in Type 1 diabetes care.
- Python
- LLM
- RL
- Program Synthesis
- 02
iOS ApplicationIslet Diabetes
Personal diabetes management for iOS — syncs Apple Health, tracks glucose, medication and meals, with GPT-4o mini for personalised insights.
- Swift
- iOS
- GPT-4o
- Apple Health
- 03
Research CodeLSTM_BGF
PyTorch implementation of a Long Short-Term Memory network for blood-glucose forecasting.
- Python
- PyTorch
- LSTM
- 04
Research CodexLSTM_BGF
Extended LSTM architecture with attention and residual connections for improved glucose prediction.
- Python
- PyTorch
- xLSTM
- Attention
Index
- Occupation
- AI & Machine Learning Engineer, Dexcom R&D.
- Education
- BSc (Hons) Artificial Intelligence & Computer Science, University of Edinburgh, 2025.
- Focus
- Machine learning for healthcare — glucose forecasting, interpretable decision support, diabetes technology.
- Location
- United Kingdom.