Machine Learning and Artificial Intelligence (AI) in Biotechnology – We are in a golden age of medical research
Written by Saptarshi Sengupta, Sr. Product Marketing Manager – Denodo
Biotech, medical research and drug discovery – all of these come at a very high cost but brings along significant benefit to humankind. A blockbuster drug can cure a critical disease for hundreds of thousands of patients across the globe and can earn the pharma company making the drug billions of dollars in revenue. That is why, just to bring one of the blockbuster drug to market, companies spend hundreds of millions of dollars, if not billions and spend decades in research, not knowing where the research will bear fruit or not. That research landscape is changing. Now is the age of genomics and genome specific drug discovery. Venture firm and medical research firms are pouring money and resource in genome sequencing and analysis. If you think about genome sequencing and analysis, at the heart of it is immense computational power, aka supercomputing. Companies like IBM, NVidia, and others are also attracting the big investment dollars to build exascale supercomputers.
Is it software alone that is changing the medical research industry?
Mark Andreessen once famously said “Software is eating the world”. Software and hi-tech industry in general has had significant impact on biotech industry as much as on other industries such as automotive, manufacturing and many others. But it is more than just the software. It is a combination of software and hardware that work in tandem to make the exascale computing possible. Recently in a MIT Technology Review, NVidia CEO Jensen Huang said that “Software is eating the world, but AI is going to eat software”. Technology companies and many Silicon Valley investors are now pouring money into artificial intelligence. But where we stand today, artificial intelligence in most of the cases mean machine learning or deep learning.
Let us make sure we understand the subtleties of how artificial intelligence and machine learning are not the exact same thing. Even though the terms artificial intelligence and machine learning are frequently used interchangeably, machine learning is just a specific application of AI whereby machines are given access to a trove of data through which machines are expected to learn by themselves.
Machine learning in biotech involves collaboration
When it comes to biomedical research, data is normally spread across many different entities across the globe. Some of the major entities include universities carrying out primary research data, biobanks holding biomedical sample and sample data, pharmaceutical companies holding drug data and biotech companies holding patient data among many other things. It is a complex ecosystem of various entities spread across the globe. Collaboration among these entities, including innovative partnership models, customer engagement and trust in data is of paramount importance. Creating the right platform and engagement model means choosing the right technologies. Cloud based bioinformatics and big data based bio informatics are necessary ingredients to create such a scientific environment which facilitates data processing and data access to all stakeholders.
Necessity of a flexible and agile platform
Machine learning or deep learning involves not only understanding complex genome sequence, cell structure and organ structure but patient demographics, drug interaction with affected cells, other external environmental factors, just to name a few. It is not only difficult to perform machine learning based on standalone big data platforms or cloud platforms but almost impossible. A true machine learning platform can be created with a right fabric as the foundation. Data virtualization is one of the popular choice for such a platform that has started catching steam in recent times. BioStorage ISIDOR platform is a great example of such a fabric created with data virtualization at the core of the platform.
Brooks Life Sciences acquired BioStorage and its ISIDOR platform for biological sample lifecycle management. The ISIDOR platform consists of BioInventory, a secure cloud-based, Web accessible storage system for global sample inventory data; BioConnect, a customized approach to data virtualization/data integration of sample inventory and research data from any location around the world; and BioInsight, an intelligent visualization and reporting tool connecting sample and research data to deliver valuable business insight. The Denodo Platform for data virtualization is the middleware in the BioConnect component of the ISIDOR platform. As part of Brooks Life Sciences, the platform also offers BioStudies, a source of clinical data for academia and research institutions, and BioMonitor, which monitors the temperature of storage units.
Beyond machine learning and AI – parting thoughts
While we are leaping ahead at the speed of light towards a better future in healthcare with advancements in biotechnology, mounting concerns around privacy, data protection and governance prevails and may become worse in the foreseeable future. While creating your own bioinformatics platform, you and your organization should keep in mind the legal implications of PII protection and misuse of genome sequencing. With proper privacy, security and governance in place, a bioinformatics platform created using big data, cloud, exascale computing or other modern data platforms can bring in a new era in healthcare system.