͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ   ͏ β€Œ  ­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­ Β­

Theodore’s Heart Initiative Newsletter:

Third Edition

Research and Experiments

On Our Horizon:

Combining Models Into the Web Application

In conjunction with the current series of experiments targeting a more general anomaly detection process along with the web application developed for the Kiro Hackathon last month we are looking forward to updating the application with a streamlined approach that should both make it easier to use and more reliable as a proof of concept for what we are trying to achieve.


Our current plan is to combine the newly optimized weights for the general model from experiment 4 with the segmentation model trained on the the same dataset to run in parallel and provided weighted results for any submitted image with each result helping to add more context to the output of the other model. In this way we should be able to provide a much more consistent and reliable assessment of images containing some or all four chambers of the heart.


Since we are combining two more powerful models into one flow we will be updating the user interface in the application to no longer prompt the user to select a model. Now the user will simply upload their image and get a result along with the results that were used to arrive at that classification.


We are excited to share the updates to the platform on our main site as soon as they are ready and hope that it will serve as a great example for what we are trying to accomplish when having conversations with medical providers and other researchers.


We are so incredibly touched by all of the generous gifts and would like to thank you all for the continued support. Our work would not be possible without the gifts we have received.

Theodore's ten perfect toes.

Issue: Posted: 9/23/2025