Data intelligence industry innovation service media
——Focus on digital intelligence and change business
On April 27, the Amazon Cloud Technology Medical and Life Sciences Industry Summit was held in Shanghai. The conference brought together industry experts and thought leaders to discuss the ways of digital transformation and innovation in the industry.
"Medical and life science" is an eternal proposition of human society, and the continuous development of science and technology provides new perspectives and solutions for modern medical research.
Since 2013, Amazon Cloud Technology has established teams in the professional medical and life science industries around the world. In the ten years from 2013 to 2023, Amazon Cloud Technology has served more than 4,200 customers in the medical and life science industries around the world, including leading companies such as Pfizer, Bayer, Roche, Merck, GE, and Philips. Nine out of the world's top ten pharmaceutical companies choose Amazon Cloud Technology. In China, Amazon Cloud Technology also serves more than 400 leading companies in the medical and life science industries.
In the long-term focus on biopharmaceuticals and the process of meeting customer needs, Amazon Cloud Technology found that the current and future medical and life science industries are facing three major challenges:
First, the surge in data volume brings insight challenges. Scientists have discovered that the number of molecules that may become drugs in nature is 10 to the 60th power, while the number of atoms in the solar system is only 10 to the 50th power. The surge in the amount of data will bring about an exponential increase in the difficulty of R&D, while slowing down the speed of R&D and increasing the cost of R&D.
Second, the challenge of continuous and rapid computing power demand. Gene sequencing technology may take 13 years to complete the gene sequencing of one person in 1990, but now, it can complete the gene sequencing of 60 people in one day, and the cost of gene sequencing per person has dropped to one million in the past 90s one-third. Behind the large-scale gene sequencing, tens of thousands of virtual servers need to be called at the same time. On the other hand, the development of computing power provides the basis for the explosion of generative AI. All of these need to call tens of thousands of virtual servers at the same time, which is a challenge to continuous and rapid computing power.
Third, customers need solutions rather than a single technical service. The core appeal of customers is that they can use "out-of-the-box" solutions to solve application problems without having a professional computer background. For example, there are nearly 100 types of medical information software used in hospitals, and about 20 are commonly used. Medical institutions need solutions suitable for industry users, and they can use out-of-the-box solutions to support business without professional computer background.
How does Amazon Cloud Technology respond to the three major challenges of data, computing power, and solutions?
Gene sequencing data full lifecycle management solution to cope with surge data
The surge in data has brought about two major needs of users for data storage and rapid automated data analysis.
According to the estimates of "Nature" magazine, by 2025, more than 60 million people in the world will use gene sequencing to diagnose diseases. The amount of whole genome data of a person usually exceeds 50GB, and the growth of genetic data in 2025 will exceed 40EB (1 EB=1024GB*1024*1024).
Therefore, how to realize safe, efficient and low-cost storage of data is an urgent problem to be solved.
For rapid and automated data analysis, you can imagine a game of "find the difference". When two pictures are compared together, we can easily find the similarities, differences and connections. However, when the number of pictures reaches hundreds of millions, human recognition has reached its limit. Relying on computer automated data analysis is an inevitable solution. Otherwise, no matter how much data is stored without analysis, medical data will not be of any help to the business.
Faced with the surge of data, Amazon Cloud Technology can provide customers with guidance on the full life cycle management of gene sequencing data, and provide better cost performance and higher availability for the storage, retrieval and analysis of bioinformatics data at different stages of gene analysis.
Specifically, the genomics data is uploaded to Amazon Simple Storage Service (Amazon S3) and managed; the genomics data is analyzed and processed using the analysis service on the cloud, and the data can be read directly from Amazon S3 and processed after analysis. Deposit results back; use Amazon S3's access control features to control access to data and share data with collaborating researchers; and regularly back up genomics data to keep data safe and archive data that no longer requires frequent access to Amazon S3 Glacier or Amazon S3 Glacier Deep Archive to reduce long-term storage costs.
At the re:Invent Global Conference in 2022, Amazon Cloud Technology released a very important managed service, Amazon Omics (genome data analysis). Amazon Omics can not only intelligently store genetic data and multi-omics data in Amazon S3 (object storage service), but also has two built-in open source gene sequencing analysis algorithms. At the same time, Amazon Omics has been closely integrated with the data analysis service on the entire cloud platform of Amazon Cloud Technology and the artificial intelligence service Amazon SageMaker. Customers can use Amazon Omics to complete the storage and analysis of one-stop genetic data, and use machine learning to integrate personal genetic data. The combination of data and more dimensional clinical data provides personalized services for more individuals.
Take Unknown Jun Biological Company as an example. Unknown Jun is an AI pharmaceutical company that focuses on intestinal microecology. In the field of microorganisms, a gene sequencing sample will consume 100G to 200G of storage capacity, and with the increase of gene sequencing projects, each project will generate massive data, and the pressure on data storage costs is huge. Through the guidance of the automated data lifecycle management solution provided by Amazon Cloud Technology, Unknown Biotech stores different data in different levels and archives them to lower-cost storage layers, which not only effectively reduces storage costs, but also has the opportunity to better improve analysis s efficiency.
High-performance computing platform on the cloud, speeding up the research and development of innovative drugs
The development of innovative drugs is often accompanied by long cycles, high costs, and high risks. Therefore, there is often a saying "121" in the field of new drug research and development, that is, a new drug often takes 10 years to develop and costs 2 billion U.S. dollars, but only has a 10% success rate. . It can be described as "a narrow escape". In order to improve the efficiency of new drug research and development and increase the success rate, modern innovative drug research and development often uses new technologies such as high-performance computing, machine learning, and quantum computing to explore, and these are also important ways to solve computing power challenges.
There is a key step in the process of computer drug development, called virtual screening, which is to verify whether certain compounds have the opportunity to become drugs through the binding analysis of known compounds and viral proteins (targets).
In general, scientists need to screen 1 billion compounds for simulated binding to target proteins. A single-core server would take 475 years to complete, but on the Amazon cloud technology platform, the virtual screening of 1 billion compounds can be completed within 24 hours. This is the advantage of using a high-performance computing platform on the cloud to schedule large-scale computing resources.
Amazon Cloud Technology has created a series of specially optimized hosting services for high-performance computing applications on the cloud to accelerate the development of new drugs. For example, EC2 computing instance optimized for high-performance computing; EFA high-performance network up to 400Gbps; FSx for Luster system that supports millisecond-level transmission; Amazon PrarllelCluster, a management tool for cluster scheduling, and support for tens of thousands of concurrent processing tasks Amazon Batch.
In addition, Amazon Cloud Technology can also provide guidance on industry solutions, allowing some customers to independently build a new drug research and development, computing power scheduling and monitoring platform. For example, protein stack prediction containerization solution guidance, SOCA open source HPC collaborative solution, etc.
Amazon Cloud Technology has also built a complete quantum computing ecosystem . The "Drug Discovery Solution for Quantum Computing Exploration" launched can provide a one-click deployment of quantum computing/classical computing hybrid architecture, and use quantum computing resources to conduct experiments through the Amazon Braket quantum computing platform. , and can be compared with calling classical computing resources, and also provides visual reports to explore new ideas for the application of quantum computing in drug discovery.
Jingtai Technology is a well-known drug research and development unicorn in China. It has built a drug screening platform based on high-performance computing on Amazon Cloud Technology, which has greatly shortened the time of new drug development and can save a lot of operating costs. . Amazon cloud technology can not only meet the resource and cost requirements of Jingtai's business, but also can call large-scale supercomputing resources at any time without purchasing large-scale clusters required by the business, and can also provide excellent computing cost performance by providing bidding instances . Liu Yang, CTO of Jingtai Technology, said in an interview, "With the help of Amazon Cloud Technology, we can not only quickly build a flexible, scalable, and easy-to-manage high-performance computing cluster on the cloud platform, but also very importantly, Jingtai The technology uses Amazon's very unique technology, EC2 SPOT bidding instance, and the cost can be saved by 50%-60%."
Quickly build innovative solutions to meet industry user experience
For users in the medical and life science industries, they not only need cloud services, but also more solutions, especially out-of-the-box solutions that meet industry requirements and customer needs, rather than building from scratch.
Based on customer needs, Amazon Cloud Technology can provide solutions covering the entire value chain of biomedicine, which can help customers quickly and efficiently conduct research and development, testing, manufacturing, commercialization and subsequent use monitoring of various drugs and medical equipment. Meet stringent compliance requirements.
Amazon Cloud Technology divides customers into builders and purchasers. Builders hope to quickly build their own solutions based on Amazon Cloud Technology's technology; buyers want to focus on business and often purchase partners directly without building their own platforms. solution. According to the needs of customers, Amazon Cloud Technology and its partners provide industry solutions suitable for the personalized experience of builders, users, and managers. These solutions include solutions from a large number of partners of Amazon Cloud Technology, such as Sales Scilligence's laboratory notebooks, Fastone's out-of-the-box high-performance computing drug development solutions, and consulting partners such as Tenthpin, Deloitte and other consulting partners to create life science work based on Amazon cloud technology load solution. These solutions comprehensively cover the whole process value chain of biopharmaceuticals, including research design, clinical trials, manufacturing, marketing promotion, post-marketing monitoring and support.
In the process of deconstructing users in the Amazon cloud technology service industry, Gu Fan, general manager of the Strategic Business Development Department of Amazon Cloud Technology Greater China, cited a case of partner programs. Yitikang (Beijing) Technology Co., Ltd. is one of the most professional smart remote ECG platforms and professional consultation service providers in China. Yitikang is a machine learning service based on Amazon cloud technology, creating an intelligent remote ECG consultation platform. In January 2020, when the new crown pneumonia epidemic just started, the remote consultation mini-program served more than 3,600 remote ECG consultations in small and medium-sized medical institutions within 30 days, which greatly relieved the pressure of diagnosis and treatment in large hospitals and reduced the flow of personnel across regions. Yitikang's solution is built on the machine learning service Amazon SageMaker, which has doubled the efficiency of AI model training and reasoning, and greatly shortened the time to market. It helps partners to complete the whole process from training to deployment and launch of the AI model that originally took half a year to complete in about 3 months, accelerating application iteration.
Empower industry customers in an all-round way to create an ecological chain
For users in the empowering medical and life science industries, in addition to solving the three major challenges of data, computing power, and out-of-the-box solutions, security compliance and how to make better use of generative AI technology are the future of Amazon cloud technology. challenges to be faced.
In this regard, Amazon cloud technology has introduced four important innovations:
The first is Amazon Bedrock. Customers can access Amazon Cloud Technology's own large-scale model Amazon Titan through the API, as well as a third-party pre-trained LLM basic model.
The second is that Amazon released two large-scale generative language models of Amazon Titan, one is text generation, and the other is digitization of words.
The third is infrastructure. For generative AI, in addition to computing power scale, cost performance is more important. The reasoning and training chips self-developed by Amazon Cloud Technology, based on the instances of these two chips, Amazon EC2 Trn1n and Amazon EC2 Inf2 are officially available, creating the most cost-effective generative AI infrastructure.
The fourth is Amazon CodeWhisperer, an AI programming assistant that generates code suggestions in real time.
In addition to technology and solutions, Amazon Cloud Technology is also working very hard to build an ecological chain of digital innovation in cloud medical and life health industries. For example, IDAC, Shanghai Amazon Cloud Technology Life and Health Digital Empowerment Center, empowers our enterprises through several aspects such as the Excellence Exhibition Center, Intelligent Network Cloud Platform, Industry Club, and Global Cooperation Plan.
Gu Fan said: "We have always emphasized that compliance is the cornerstone of everything, and the emergence of generative AI is worth exploring its application scenarios in the medical and life science industries with Amazon Cloud Technology. At present, Amazon Cloud Technology not only provides coverage Global cloud infrastructure and more than 200 categories of cloud services. More importantly, we have a deep understanding of industry needs, continuously enrich and expand the industry's digital innovation ecological chain, and cooperate with partners around the needs of data, computing power, and experience. Together, we have launched many end-to-end solutions for customers that conform to the development trend of the industry."
Text: Mu Yang / Data Ape
