Researchers collected data from 23 patients with Alzheimer's disease: brain scans, blood tests, and microbiome composition. Each patient underwent measurements before and after treatment with rapamycin, a drug that slows cellular aging. The neural network's job is to figure out from these data whether a person received the drug. But the signal is weak and drowns in noise, like a grain of gold in river sand.
First, the model "sifts the sand": a random forest selects the 27 most important features out of hundreds. Then the neural network matches data of different types and predicts treatment. Without sifting, accuracy was random, around 50%. With sifting, it rose to 71.25%. Blood parameters and the inflammatory marker MIP-1α turned out to be the most important. Now it's clear: to catch a drug's effect, you first need to remove the noise. This approach could already help doctors select patients whom rapamycin is most likely to help.
Rapamycin was first isolated from a bacterium found in the soil of Easter Island (Rapa Nui), hence the name. The drug suppresses the mTOR signaling pathway, which regulates cell growth and aging, and extends the lifespan of mice by 10-15%.
The idea of predicting how a body will respond to a drug from a multitude of biological signals echoes Frank Herbert's novel Dune, where doctors use complex diagnostic systems to assess a person's condition. Unlike science fiction, though, this is about real blood and brain data.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.