There is a well-known "double ten law" in the field of new drug research and development: it takes an average of 10 years and US $1billion to develop a new drug. Is there a more efficient way? With the development and progress of artificial intelligence, the importance of data for drug design, discovery, clinical validation and other links has become increasingly prominent. However, compared with the whole industry, any pharmaceutical enterprise or R & D institution has a very limited amount of data. For competitive reasons, they can only rely on their own data for drug research and development. The amount of data has become one of the key factors restricting the success or failure of artificial intelligence drug research and development. Recently, a data platform involving 10 top pharmaceutical companies in the world and 17 partners was announced. On this platform, enterprises competing with each other can obtain calculation results without exposing the data they have. This artificial intelligence new drug research and development project aims to break the "data island" between different subjects and explore a new mode of data sharing - using the data of multiple pharmaceutical enterprises to create more accurate models and screen the most effective compounds for drug development. The project brings together 10 global top pharmaceutical enterprises, including Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, GlaxoSmithKline, Janssen Pharmaceutical, Merck and Novartis. The partners include an artificial intelligence enterprise, four start-ups and two European universities. Among them, owkin, a privacy computing company, provides a blockchain based data model platform, and the central scheduler allows each pharmaceutical enterprise to share the same federal learning model. In short, pharmaceutical enterprises can call more data to calculate in the model without disclosing their own data. The massive data greatly improves the accuracy and applicability of model prediction. Previously, owkin published a paper entitled "mesothelioma classification based on deep learning can improve the prediction of patient outcomes" in the journal Nature Medicine. The paper analyzed the digital biopsy images of nearly 3000 patients, and the data came from several French institutions. This time, 10 pharmaceutical companies promised that they would invest unprecedented amounts of data after the security and privacy protection of the project were proved. It can be predicted that if this data sharing mode is proved to be feasible, more data holders related to drug research will share data through this mode, and the "data island" will no longer exist. In addition to drug research and development, privacy computing will also play an increasingly important role in urban governance. Last year, Zhengzhou suffered a once-in-a-century rainstorm. Since then, local governments have attached great importance to the treatment of waterlogging and ponding. How to improve the accuracy of waterlogging prediction and turn post disaster response into pre disaster warning? Zhangjiachen, founder of the first batch of leading enterprise of privacy computing in Shanghai data exchange, said that in the past, some sensitive but practical data could not be used, such as the data of water pipe network. However, through privacy computing technology, it can be summarized together with terrain, geography, astronomy, hydrology and other weather related data to train a more accurate prediction model, And integrated into the emergency management solution of the whole water affairs system. In Wuhan East Lake High Tech Development Zone, Shanghai blockchain enterprise zero one universe is participating in building the largest "digital base" in China, including private computing. Compared with the traditional digital city, the "digital base" built by using blockchain and privacy computing technology has self-learning ability. When the algorithm of a node evolves, it can be self-learning and promotion between edge nodes. The integrated system design reserves an interface for the subsequent scenarios. In the past, the data formats and structures of different departments were different, so it was very inconvenient to interact. Shangguanyun, the executive president of zero one universe (Shanghai) Technology Co., Ltd., said that the upgrading of digital area construction from "strip and block" to "base" is to break the situation that data of various departments were not connected in the past and realize unified planning and interface standards for data. In order to let all departments share data with confidence, privacy computing must be used. It is reported that the "pedestal" digital area will greatly reduce the operating cost of the digital system. It is estimated that at least 1/3 of equipment investment can be reduced by completing the same task. In addition, in the construction of this "digital base", the "carbon ledger" technology based on blockchain and privacy computing has been preliminarily applied. Shangguanyun introduced that the "urban zero carbon" plan is divided into two steps: one is recording, the other is forecasting. Blockchain technology can record data information from the collection source, calculate certificates, and predict the carbon emissions that will be generated through the model. At the same time, not all data information needs to be disclosed. Based on privacy calculation, only the final emission results can be displayed, providing a quantitative action plan for energy conservation and emission reduction. (Xinhua News Agency)