Our Transport, Health and Urban Design Research Lab @unimelb and @msdsocial is very pleased to be recipients of a significant @nhmrc research grant to investigate urban design and health outcomes through computer vision techniques. More #spaceforhealth coming to a place near you!

To summarise the research: We have recognised the GIS and health survey techniques have underpinned much of the work in Urban design and health up to now. This has got us a long way, but the methods are limited by scale. It's really time consuming and expensive to ramp it up.

But the availability of international image datasets containing maps, satellite images, and street-view images means we don't have to limmit ourselves to GIS data, anymore, we can go straight to the images and treat them as data. We can see roads, buildings, bike lanes, paths...

We can collect these images very efficiently from all over the world, then use powerful analysis techniques drawn from AI to group the images, and cities into various types at whatever scale we want.

Adding global health datasets into the mix, we can then associate these inter and intra-urban typologies with population health outcomes: heart disease, diabetes, respiratory disease, infectious disease (dare I say it, COVID-19 for example). Each city has it's own 'fingerprint'.

We can then provide this information to cities, show them which areas of the city are problematic, and given them examples from other similar cities that have made improvements. We can also use GANs to show how the style of cities can be altered to make them healthier.

A big thanks to the project and research team @msdsocial, Prof Mark Stevenson, Dr Jasper Wijnands @jasper_seb and Dr Kerry Nice @mothlight assisting to pull this together. Epidemiology, maths, computer science, psychology, and geography. Multidisciplinary teams are great teams!

And most importantly, apologies for all the spelling mistakes in this thread :) It's 2 degrees here and my hands are cold!