Philip Fradkin

Title: Philip Fradkin and Molecular-Phenomic Machine Learning

Introduction

Philip Fradkin is an inventor associated with Recursion Pharmaceuticals, Inc. in Toronto, Canada. His patent work concerns machine-learning methods for connecting molecular structures with phenomic images to help characterize how molecules affect cellular functions.

Latest Patents

Fradkin is named on one patent, titled “Utilizing Contrastive Machine Learning Models to Extract Joint-Space Molecular-Phenomic Embeddings from Molecular Structures or Phenomic Images.” The disclosed systems and methods use a contrastive model to learn a shared latent space for molecular structures and phenomic images. The approach incorporates phenomic image embeddings from a pretrained encoder, an inter-sample similarity-aware loss, concentration encoding, and undersampling of inactive molecular data. The resulting embeddings can support comparisons of molecules and images, predictions of phenotypic impacts, and classifications of molecular activity. Fradkin’s listed portfolio includes 1 patent.

Career Highlights

Fradkin’s listed company affiliation is Recursion Pharmaceuticals, Inc. His patent reflects work at the intersection of machine learning, molecular data, and cellular phenomics, with potential applications in analyzing molecular activity and biological effects.

Collaborations

Jan Frederik Wenkel is identified as Fradkin’s coworker. The patent’s disclosed approach combines molecular structures, phenomic images, and machine-learning techniques to produce representations useful for molecular inference.

Conclusion

Philip Fradkin’s patent contribution focuses on using contrastive machine learning to connect molecular structures with phenomic images, enabling computational analysis of molecular effects on cellular functions.