You made it to the second episode - excellent!

Part 2 on Health data poverty. This is an example of a team actually doing something about some of the issues we covered in Part 1 with Dr Xiao Liu.

Delighted to have spoken to Prof Alexandre Filho, Professor of Machine Learning in Sao Paolo, Brasil who told of the work him and his team have done in maximising the impact of data driven technology in Brasil.
"The world is becoming more like Brasil"
Aside from carnivals and football, what else can we learn from our colleagues in Brasil looking to improve things for underserved communities with data diversity?

We covered:
- What the world can learn from this work and Brasil's high quality and already diverse data
- The challenges in data availability and limitations in performance of algorithms developed in richer settings (even in the same city in Brasil) that are deployed rurally.
- What work they have done to overcome this: Leverage data you have to build inclusive datasets : ⁠This paper about the work the team did⁠ on neonatal mortality prediction with routinely collected data.
- In addition the people who the data was collected from actually benefitted from the work.
- The trade off between being tuned and high impact to one context as the data is hyper relevant, and the need to scale and be able to generalise to wider contexts - otherwise we have too many expensive local solutions.
- The extended work to create benchmarking in the ITU/WHO focus group AI for health to evaluate performance of similar models with routinely collected data in other LMIC settings - what are the core overlaps, and what are the important contextual differences?
- Transfer learning and generalising models to better performance and impact locally.
- Prof Alex's key recommendations to innovators and implementers in the EU/US/UK.

Alexandre Chiavegatto Filho is an Associate Professor of machine learning in healthcare at the Department of Epidemiology of the School of Public Health, University of São Paulo. He is the director of the Laboratory of Big Data and Predictive Analysis in Health (Labdaps) at the University of São Paulo, which currently has a team of 30 researchers focused on developing artificial intelligence (machine learning) algorithms to improve healthcare decisions.
Find the team's work on ⁠Google Scholar⁠