Invited lecture ·
Enabling Personalised Care using Real-world Machine Learning in Multiple Sclerosis
A CUBRIC seminar on biological MS types, automated MRI measurement and federated analysis across hospitals.
CUBRIC Seminar · Cardiff, UK
- Delivered
- Event
- CUBRIC Seminar
- Place
- Cardiff, UK
- Format
- Invited lecture
- Source deck
- 22 slides
From the slides
Deck notes
This seminar connected biological MS types with individual treatment effects and the practical work needed to analyse routine MRI across hospitals. It also introduced automated regional brain measurements, SynthSpine and the wider MS-PINPOINT programme.
The question
How can biological disease patterns support personalised care while hospital data remain within their clinical settings?
Deck outline
- Data-driven MS types derived from MRI
- Individual treatment effects and treatment risk–benefit
- Automated measurement of lesions, grey matter and ventricular regions
- SynthSpine for spinal-cord MRI data augmentation
- Federated learning across participating hospitals
- Translation through the MS-PINPOINT programme
Take-away
Personalisation depends on robust measurements and deployable, privacy-preserving infrastructure as much as it does on the model itself.
Original deck
Slide by slide
Read the 22-slide presentation in order, or use the index to jump.
Slide 01
Enabling Personalised Care using Real-world
Slide 02
Outline
Slide 03
Jacqueline Du Pré
Slide 04
Eshaghi et al, Nature Communication, 2021
Slide 05
Research software
Slide 06
Slide 6
Slide 07
Slide 7
Slide 08
Goebl, …, Eshaghi, Nature Communication, 2025 (in press)
Slide 09
Goebl et al, Nature Communications (2025), in press
Slide 10
Goebl et al, Nature Communications (2025), in press
Slide 11
SynthSpine: Spinal cord MRI data augmentation framework
Slide 12
Slide 12
Slide 13
(2022 to present)
Slide 14
MS-PINPOINT project
Slide 15
Federated learning
Slide 16
Slide 16
Slide 17
Cardiff Cohort
Slide 18
Radiology reports (~6,000)
Slide 19
Slide 19
Slide 20
Slide 20
Slide 21
CVUHB/Cardiff University
Slide 22
MS-PINPOINT group
