Ana Matoso

Can we? (yes) Should we? (no) Did we? (nobody really knows)

And you can't visit Honolulu without doing 2 things: catching a lot of sunsets, and doing the Diamond Head trail (171m elevation, avg 8.6% incline)- which, with the "warm-up" of doing the Lulumahu Falls trail (avg 20% incline) before, was quite easy and worth it for the view.

Last week, I attended the @ISMRM Annual Meeting in beautiful Honolulu, Hawai'i, with my colleagues from @laseeb_isr. There, I presented a pitch and a poster on my PhD work titled "Prediction of treatment response in a longitudinal glioblastoma dataset using deep learning".

Deep learning can help with glioblastoma prognosis. Ask your questions to @anamat050 at her power pitch and digital poster "Prediction of treatment response in a longitudinal glioblastoma dataset using deep learning". 13:50, Theater 1, computer 19.

🧵 3/3 Perhaps it's time we reassess our models' target metrics so that they are more aligned with the reality of medical practice, where reliability is just as important if not more, than accuracy. It all comes down to trust.

🧵 2/3 It cannot be understated that robustness is crucial for risk assessment and for comparing different solutions, especially in a medical context. As such, it should be reported on. However, we need to be wary because significance can be easily achieved with big enough data.

Very nice paper on how, in the race for SoTA results, it appears that we have fallen into a peculiar trap: only reporting the peak/average performance thus overlooking the variability of the results. arxiv.org/abs/2409.17763 🧵 1/3

4/4 🧵 - More on phronesis "Understanding what it is to live well with others and what we owe to one another — that conversation can’t ever stop. And so I’m very reluctant to pursue the development of machines that are designed to find an optimal answer and stop there"

🧵3/4 - On practical wisdom (phronesis) "I think if we had a truly moral machine (...) It would be able to articulate reasons and appreciate our reasons, and negotiate those reasons with us in a way that wasn’t just mirroring the consensus that we’d already reached."

Great interview on how we perceive AI and on the morality of using it for decision-making. As an engineer, I find it important (and refreshing) to listen to philosophers talking about this to see it from a different lens Some favourite quotes in 🧵 vox.com/future-perfect…

New from Wei, Poon, and Wang--BiomedParse is a biomedical foundation model that can jointly conduct segmentation, detection and recognition across nine imaging modalities. nature.com/articles/s4159…

Very insightful article on how judges selectively ignore outcomes of an AI algorithm unless it aligns with their decision Funny how they try to eliminate bias by adding a biased algorithm to a biased process, only for it to suffer from confirmation bias shorturl.at/WK91j

Congratulations @sara_pmonteiro! It's incredible to see that by using your new experimental setup, such amazing results can be found. Keep up the good work!🚀💪

They really meant it when they called it sLOVEnia because I loved it, the people, the place and of course, the conference! Looking forward to the next one! I'll leave you with some photos of the natural beauty I found 3/3🧵

I went with two @laseeb_isr colleagues: Catarina Passarinho, who also gave a talk at the conference, and Marta Loureiro, who presented a poster and won a best poster prize! 2/3 🧵

Just got back from Slovenia for the 1st Net4Brain COST action meeting where I presented the first results of my PhD. It is key that researchers from several fields get together and discuss their different viewpoints and ideas to beat brain cancer. 1/3🧵

@Dee_Kay_Jay Ticketmaster said mayybeeee 😂 Don't lose hope! There's always people who then will want to sell them. 🤞

Após quase 4 anos, já poupei quase tanto como ouvi de podcasts. Sim, eu oiço a 2x e corto os silêncios no extremo

Finalmente posso avaliar podcasts! Eu e as mais 5 pessoas que usam esta underrated app em Portugal. Já fui a primeira avaliadora de vários podcasts portugueses. Experimentem, prometo que nao se arrependerão