Federated AI Learning

Overview

A distributed machine learning approach where organizations contribute to shared model improvement without exposing proprietary data, proposed as the future model for CDMOs and cross-industry biopharma collaboration. Federated AI learning with secure data sharing (potentially blockchain-enabled) is proposed as the future operating model for CDMOs, allowing them to improve bioprocessing models by accessing collective knowledge without revealing confidential process data. Merck piloted cross-industry data sharing for predictive maintenance model training, sharing non-sensitive equipment data with partners to improve model accuracy, resulting in near-zero equipment failure modes. (Segment 9 (40:00-44:37))