Reza Sameni

Glad to announce a $50,000 Data Challenge on ECG image digitization, organized by our PhysioNet Challenge team and Kaggle, expanding on the 2024 G.B. Moody PhysioNet Challenge on ECG image digitization. Learn more and participate: kaggle.com/competitions/p… #ECG #Kaggle #PhysioNet

We are excited to announce the opening of the George B. Moody PhysioNet Challenge 2025. This year, the challenge focuses on developing algorithms to identify Chagas disease using electrocardiograms (ECGs). Chagas disease affects over 6.5 million people worldwide, with a significantly higher prevalence in the Global South. It is often detected through serological testing, which is limited in availability. Using ECGs for automated detection can help prioritize testing and improve treatment efforts. Participants will work with initial training data from the CODE-15% dataset, with additional data to be shared as the challenge progresses. Example entries and scoring code are available in both MATLAB and Python to help teams get started. As usual, the challenge is divided into two stages: an unofficial phase for community feedback and an official phase. This year's challenge will conclude in Brazil at the annual Computing in Cardiology conference, where winners will be announced. Additional prizes are available to encourage participation from teams in the Global South. Learn more and register at lnkd.in/e9vK9MBH.

@rahallclifford Congratulations Rachel!

🎉Tenured! Again!🎉 I promise to use my powers for good trouble. And I'm SO grateful to the many colleagues and mentors who made my move to Emory possible!!

@JLucasMcKay Excellent work @JLucasMcKay!

Excited to announce the George B. Moody PhysioNet Challenge 2024, focused on digitizing and classifying electrocardiograms (ECGs) from images and scanned papers! Read more and join the unofficial phase today: lnkd.in/ejQ9MVMb @emorybmi @CinC_tweets

@zhang_yueting @gabrielpeyre Indeed, a linear systems equivalent to the central limit theorem is that the cascade of many linear systems bahaves like a single Gaussian-shaped impulse response, which is also Gaussian in the frequency domain :)

The product of two Gaussians is a Gaussian. Sums of Gaussians is an algebra for product. It is is the hump algebra, studied by Yves Meyer (Gauss prize 2010, Abel prize 2017). It is a Besov space. cambridge.org/core/books/wav…

@StephaneDeny @MBCarr0 @gabrielpeyre Yes - we just need robust measures for gaussianity, which is basically what the different flavors of classical ICA algorithms boil down to. Of course, there's a different approach based on the joint approximate diagonalization of eigenmatrices, like JADE and SOBI. Fun stuff!

@RezaSameni @MBCarr0 @gabrielpeyre I know this in the limit of a large number of independent variables (central limit theorem), but is there also a similar theorem for say two variables?

Oldies but Goldies: Pierre Comon, Independent component analysis, a new concept?, 1994. Gives a rigorous definition of the ICA as an optimization problem, leveraging non-Gaussianity and independence of the sources. en.wikipedia.org/wiki/Independe…

Join this discussion on the challenges of building #AI models from electrocardiogram data on this “ECG Segment: Making Waves” episode of the “Mayo Clinic Cardiovascular CME” podcast with host @anthonykashoumd and guests @GariClifford and @RezaSameni. mayocl.in/3t5QRx3

Thrilled to share that @EmoryUniversity's safe+natal program, in partnership with the Maya Health Alliance | Wuqu' Kawoq is among the 15 recipients of the @Googleorg's AI for the #GlobalGoals Impact Challenge grant. @GariClifford @rahallclifford @Nasimkatebi @emorybmi

Thanks to the community support, Passive-Brain-Audition is now among the top 32 #ARPAHDash Bracket ideas. Read more and vote for the second round here: arpa-h.gov/news/dash/ #ployplexus #ARPAH #emoryuniversity

Our *Passive-Brain-Audition* idea is selected among the #ARPAHDash ideas. We proposed to develop an AI technology for early detection of brain aging and dementia from brain wave responses to environmental and memorable sounds during daily life. Please vote if you like the idea!

@martinmbauer Very interesting. What's the interpretation of this curve? Since R, volume and surface areas have dimensions, each point of the curve has a different dimension and we can't compare them without unit normalization. Or maybe I'm missing something.

Similarly for R=1 the n-volume of the n-sphere has a maximum around d=5. Dimension 5 has the fattest unit sphere. In the limit n -> infinity the unit sphere has no surface and no volume. 5/11

A short thread on higher dimensional spheres. Or what the volume of spheres have to do with (-1)^-i ? A sphere is the collection of points with equal distance to it's centre. 1/11