31. Dragonfly Explores Titan, Curing Muscular Dystrophy, NeckFace

Engineering & simulationconvolutional neural network50-fold increase in penetration

The AI model analyzed experimental data on cell-penetrating peptides to identify optimal amino acid combinations, resulting in a mini-protein that increased drug penetration 50-fold without damaging cells.

“they created a convolutional neural network that analyzed, it was basically looking for patterns, what groups of amino acids work the best together, and it provided feedback on what we should actually be spending our time investigating.”

Engineering & simulation50 times better penetration in mice

Researchers used an AI model trained on experimental data to predict which peptide combinations would best deliver a muscular dystrophy drug, leading to a candidate that increased penetration 50 times in mice without harming them.

“They ran it again, and they got a proposition with this, the CPP, or mini protein, actually, that could increase the penetration 50 fold, but they have to test it out, right? Because the computer's telling you one thing that you want to see if it works. They test it on mice and it actually turns out to be true.”