All talks & lectures
Conference ·
Clinical prediction using complex models
A measured look at where complex models can improve multiple sclerosis prediction and where complexity can obscure understanding.
ECTRIMS 2022
- Delivered
- Event
- ECTRIMS 2022
- Format
- Conference
- Source deck
- 23 slides
From the slides
Deck notes
This ECTRIMS talk considered the role of complex models in clinical prediction. It moved from simple relationships to personalised risk, MRI-derived subtypes and emerging outcomes.
The question
Where does model complexity add clinical value, and when does it make a prediction harder to understand or use?
Deck outline
- Relationships, models and clinical questions
- Where predictive models may help
- Disease activity and phenotypic conversion
- Personalised disability risk
- MRI-derived multiple sclerosis subtypes
- Emerging outcomes and future perspectives
Take-away
Complexity is useful when it improves a defined clinical decision; the model still needs an outcome and explanation that clinicians can assess.