Deep learning for drug discovery, diagnostics and more
PFN’s life sciences research and applications include drug discovery, omics analysis, medical image analysis and compound analysis.
We are developing several technologies for drug discovery using our deep learning expertise and in-house computing resources. In a joint research project with Kyoto Pharmaceutical University, we used one of our technologies that led to the discovery of multiple lead compounds for COVID-19 treatment. We are also working on various drug discovery-related projects with Chugai Pharmaceutical including experiment automation and physical property prediction.
Omics Data Analysis
We are also looking to apply our deep learning expertise to analysis of so-called omics – genetic or molecular profiles of humans – for easy and early detection of various diseases. Preferred Medicine, our US joint venture established in 2018 with Mitsui & Co, is conducting clinical research with US medical institutions on machine learning-based breast cancer detection with blood miRNA. We are also conducting a joint research with Kao Corporation to make use of sebum RNAs collected from oil-blotting films for early detection of Parkinson’s disease as well as beauty counseling.
Medical Image Analysis
Medical image analysis is another life sciences application of PFN’s technologies. We have trained a deep learning model with a large number of actual chest X-ray images paired with lung cancer diagnosis so that it can indicate abnormalities that may represent lung cancer and facilitate physicians to diagnose the disease (left image).
In March 2021, PFN were in third place out of 1,547 teams from around the world at a Kaggle competition RANZCR CLiP in which the participants competed for classifying presence and correct placement of tubes on chest x-rays (right image).
Image credit: NIH Clinical Center. Wang X, Peng Y, Lu L, Lu Z, Bagheri M, Summers RM. ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases. IEEE CVPR 2017