SDG 3: Good Health and Well-being
SDG 9: Industry, Innovation and Infrastructure
2. Project Details
Company or Institution
A Measurements-Based Approach to Supervised Machine Learning
General description of the AI solution
Every field of science and engineering starts with measurements. To quote Siddhartha Mukherjee in `The Emperor of All Maladies`, "Science begins with counting. To understand a phenomenon, a scientist must first describe it; to describe it objectively, he must first measure it."
When working on machine learning problems, modern data science often relies more on computation (let's throw more GPUs at the problem) and guesswork (let's see if we can modify Alexnet for our specific problem) than on any kind of measurements. Brainome is an approach to supervised machine learning that is rooted in information-theoretic measurements. This approach is especially interesting for high dimensional problems that, without measurements, suffer from "the curse of dimensionality". Specifically, we had amazing success with genome decoding, for example for early Cancer detection.
The tool is available at: http://www.brainome.ai
University of California, Berkeley
Excellence and Scientific Quality: Please detail the improvements made by the nominee or the nominees’ team or yourself if your applying for the award, and why they have been a success.
Brainome is a information-theory based machine learning with Neural Networks, Random Forrests and Decision Trees. Due to the use of measurements, no hyperparameter tuning is required, results are repeatable and reproducible. The website brainome.ai contains academic publications, tutorial videos, examples, and the tool can be easily installed and tried by anybody using "pip install brainome".
Scaling of impact to SDGs: Please detail how many citizens/communities and/or researchers/businesses this has had or can have a positive impact on, including particular groups where applicable and to what extent.
The tool has an average download count of 100 downloads per day, tendency upwards.
We are working with Lawrence Berkeley National Lab, the Chen-Zuckerberg Biohub, and Cedar-Sinai hospital in LA to setup clinical trials for Brainome's output for 33 cancer types based on the Cancer Genome Atlas. The result would be a blood test for cancer using a PCR machine that would work very early. This would allow regular, early screening for cancer as part of everybody's "physical exam". Due to COVID-19 PCR machine are now widely available and so this would safe billions of dollars and countless of lives compared to the current biopsi-based testing.
Scaling of AI solution: Please detail what proof of concept or implementations can you show now in terms of its efficacy and how the solution can be scaled to provide a global impact ad how realistic that scaling is.
We have the genes responsible for 33 types of cancer, automatically detected by our measured machine learning approach. The feasibility that these results are correct has been independently evaluated using cross checks against known genes. We also have the confirmed feasibility that these are the correct genes by two different institutions. Wet experiments pending.
Furthermore, our tool is available using simple "pip install brainome" and will allow other researchers to come up with similar results, thus democratizing machine learning.
Ethical aspect: Please detail the way the solution addresses any of the main ethical aspects, including trustworthiness, bias, gender issues, etc.
Cancer affects everybody independent of gender, ethnicity, or social status. This solution is cheaper and easier to access for everybody and will increase survival rate of socially disadvantaged people.