Research
Exploring how language models and AI can support healthcare in Thailand, from clinical NLP to diagnostic decision support.
As Sheldon Cooper once said, people should know where they stand. If they don't, they need a theoretical physicist to tell them. We use models instead.
Mahidol University
Faculty of ICT
Health Informatics & Applied AI
Bridging clinical practice and machine learning
Health Informatics
Digital health systems, clinical data workflows, and interoperability standards.
Natural Language Processing
Thai medical text analysis, clinical note extraction, and named entity recognition.
Language Models
Fine‑tuning and evaluation of LLMs for clinical decision support.
AI in Medicine
Diagnostic automation, treatment recommendations, and clinical validation studies.
Astronomy & Data
Small-scale orbital observation, stellar nursery analysis, and space-based instrumentation. Hawking said to look up at the stars and not down at your feet. That's the plan.
If we can teach a model to read Thai clinical notes, we can dramatically reduce diagnostic turnaround time in under-resourced hospitals. It's not rocket science, as Sheldon would say. It's far more complicated.
Status: Concluded ✓
The real bottleneck in Thai healthcare AI isn't the model. It's the data pipeline. However difficult the problem may seem, as Hawking once wrote, there is always something you can do and succeed at. We proved that in one paper.
Status: Archived
Sometimes the most interesting data doesn't come from a hospital. It comes from somewhere much younger, much hotter, and 13.8 billion years in the making. The universe began with a bang, and the data it left behind is still waiting.
Status: Prepping ✦
"Intelligence is the ability to adapt to change."
— Stephen Hawking
I'm not crazy. My mother had me tested. No, but really, my hypotheses are peer-reviewed.
Something small is going up to look at something young. Details soon; we're still doing the math. The greatest enemy of knowledge, as Hawking put it, is not ignorance but the illusion of knowledge.
Turns out the skills that help you read messy clinical data also help when your data source is a few hundred light-years away and won't sit still. Whether it's the cosmic microwave background or a patient's chart, pattern recognition is pattern recognition.
Bazinga. Except in science, the punchline is a p-value.