Doing fun imaging genetics research @dbc_unil 🫀🧬 Former Research Scholar @aim_harvard - PhD @MaastrichtU I love coding, reading, science & a bunch of sports!
Lausanne, Switzerland
Joined July 2010
- Tweets807
- Following330
- Followers146
- Likes436
Dennis Bontempi, Ph.D. retweeted
NEW Research: FaceAge, a #DeepLearning system to estimate biological age from face photographs to improve prognostication: a model development and validation study. @HugoAerts @denbonte @Dr_RayMak
Read it here: thelancet.com/journals/landi…
New #preprint alert 🚀 (hopefully the first of many at @dbc_unil @unil 😎)
medrxiv.org/content/10.1101/…
... Did you know arteries can expand to compensate for bad cardiovascular health? A thread! 🫀🧬🧵
1
47
👉 We also investigated the ratio of cIMT to cLD, looking for a new perspective on the relationship between arterial wall thickening and lumen diameter that might lead us to a deeper understanding of arterial health!
1
23
Why should we care? 🧠✨
There is still much to learn about #cardiovascularhealth and looking at the combination of both cLD and cIMT may be better than either one alone (... more on this coming soon 👀)
#research #genetics #ultrasound #AI #UKBiobank
23
Thrilled to have helped in this😎
Congrats to the team, and many thanks to @FabianIsensee1 & @JakobWasserthal (and collaborators!) for the #OpenSoftware that made this possible! 🚀
🔬Just out in @LancetDigitalH , our work on AI for #patientsafety in #RadOnc
buff.ly/3VYptNi
Tldr
🎯 Problem: Wrong level spine treatment in medicine is a significant safety risk (counting/labeling correct vertebra is error prone, and some ppl have variant # of vertebrae).
🤖 Solution: We developed and prospectively implemented a #AI quality assurance tool into a #radiotherapy clinical workflow to check for wrong level spine targeting as an assistive technology (think lane departure assist in cars) and caught rare errors before they reached the patient, improving patient safety.
#MedTwitter #DigitalHealth #AI #Oncology @astro_org @OncoAlert @HugoAerts
47
Some great research coming out of @dbc_unil 🙌
... this one made the Nature cover as well!
By analysing the ancient genomes of individuals from Rapa Nui, researchers have overturned a contentious theory that the remote Pacific island experienced a self-inflicted population collapse before European colonization go.nature.com/3MJ98qM
55
Dennis Bontempi, Ph.D. retweeted
Contrastive learning of representations can help discover quantitative imaging biomarkers for oncology!
📝 nature.com/articles/s42256-0…
☑️ Performant in low data
☑️ Resource-efficient
☑️ Robust to inter-reader and acquisition diff.
@NatMachIntell @aim_harvard @BrighamRadOnc
🧵1/6
If you're still unsure about who to vote for the MSB Educational Challenge - I've just asked ChatGPT.
... One does not simply ignore ChatGPT.
#AcademicTwitter #MICCAI2023 #Medical #Imaging #MRI #Python #DataScience
🌟 Finalist Spotlight: Anna Zapaishchykova & @BenjaminKannMD
🎥 "Preprocessing MRI in Python"
Dive into this insightful tutorial and explore innovative approaches to MRI preprocessing.
👉 Watch now: youtu.be/2tCa7Raho6A
1
5
708
Dennis Bontempi, Ph.D. retweeted
1/ Check out this preprint that describes how we developed and validated a new #AI-based biomarker that we are calling “FaceAge”, which uses a simple face #photo to predict outcomes in #cancer patients!
medrxiv.org/content/10.1101/…
#oncology #AImedicine #FaceAge
Dennis Bontempi, Ph.D. retweeted
Google "We Have No Moat, And Neither Does OpenAI": Leaked Internal Google Document Claims Open Source AI Will Outcompete Google and OpenAI #googledocs #openai reddit.com/r/MachineLearning…
![From the MachineLearning community on Reddit: [D] Google "We Have No Moat, And Neither Does...](https://pbs.twimg.com/card_img/2121593892969578496/DopLxe69?format=jpg&name=800x419)
"Frequently in ML, researchers prefer to treat the task as a black box and let the learning algorithm figure out how to solve it. That’s fine for toy domains."
Meet Pete Wurman, Director of Sony AI America and Gran Turismo Sophy Project Lead. Pete's latest blog takes a peek under the hood of #GTSophy and shares how this breakthrough project got its start. ow.ly/Lh3F50KmhuH
1
1
Dennis Bontempi, Ph.D. retweeted
This looks like Theranos level fraud. The last 16 years of Alzheimer’s research looks to have been built on deliberately falsified data.
It’s why Alzheimer’s drugs have a 99% failure rate in trials. The thing they’re trying to fix isn’t the problem.
dailykos.com/story/2022/7/22…
GT is immensely complex and unpredictable - and it's not just time trials anymore: the people Sophy RACED against are INSANELY good.
Hats off to @SonyAI_global - this is, under so many POVs, more impressive than #AlphaStar and #AlphaGo.
... I can't wait to race Sophy on #GT7 🙃
The project behind Gran Turismo Sophy has been recognized as a breakthrough in AI and will be featured on the cover of @NaturePortfolio’s February 2022 edition.
@SonyAI_global's research behind #GTSophy was published today in Nature Portfolio. nature.com/articles/s41586-0…
1
Finally got a hold of my UCL mug we won at #MedICSS this summer @CmicUcl @WEISS_UCL!
... Pretty sure what's written on it is far from the truth, but I'll take it 👀
ALT Liquid genius mug.
Not sure our #ImagingGenetics project presentation *really* was worthy of the prize - but we certainly had the best #memes.