Medical and biotechnologies are under immense pressure. The public expects better results ever faster, but the resources are limited and can hardly be increased. This dilemma can only be solved with the most modern technology, such as AI.
Today’s requirements for medical and biotechnology are highly complex. The questions include: Can a disease epidemic be predicted and stopped before there are victims? Can we customize the treatment of chronic diseases so that no two people receive the same medicine, but equally achieve the best possible result? How can we drastically reduce the time and costs involved in drug development?
AI: An Absolute Must
To address these and many similar problems, the use of the latest technologies is essential. Today, the aim is to use automation and new methods to use the staff’s existing specialist knowledge and free their work from routine tasks. For example, artificial intelligence (AI) helps monitor patients effectively at home or better support cancer research.
A Particular Problem: Very Big Data
The successful use of AI in healthcare is based – just as in many other areas – on big data. To be more precise: to massive data! For example, a single CT or MRI examination can produce several hundred images – and a thin-slice image scan can have thousands.
This then leads to files on the order of several hundred gigabytes. So it’s no wonder that a hospital generates an average of around 665 Tbytes of data a year. Unfortunately, most of this data cannot be used directly because 80 per cent of it is unstructured, such as notes, videos, pictures or sketches. And that means: only with the help of AI can this data be evaluated so that new knowledge can be gained from it.
Applications: Image Analysis and Covid Research
These kinds of applications are now increasing in many areas of the healthcare sector. Siemens Healthineers uses AI to automate and standardize complex diagnoses. To do this, different anatomical regions must be segmented and categorized automatically.
The hardware required for this consists of the second generation of the scalable Intel Xeon processors, which, together with Intel’s Deep Learning Boost and the OpenVINO Toolkit, perform anatomical measurements almost in real-time and thus significantly increase workflow efficiency.
State-of-the-art Intel technology is also used in the current fundamental research on the new coronavirus responsible for the COVID-19 pandemic. Here, the Berlin Institute for Health (BIH) researchers are at the forefront to better understand the new virus with genetic analyses. To do this, they have to examine thousands of cells using single-cell RNA sequencing.
But such analyzes require immense computing power that the institute cannot provide itself. Intel stepped in, so the number of cells flowing through analysis could be increased by 70 per cent, which considerably speeds up the investigations.
The BHI system is based on Intel’s Select solution for the hardware configuration of genetic analyzes. The selected Dell PowerEdge R740xd rack servers are equipped with Intel Xeon Gold 6252 processors (24 cores, 20.10 GHz), P4160 SSDs and two Intel Ethernet Converged Network Adapters X710, which are operated with 10 GbE.
AI: Every Beginning is Easy
The path to new AI applications can quickly begin with existing data centres based on Intel technology. This means that there are no particular entry barriers. Instead, the first AI project can be started quickly and easily.
This is all the easier because Intel has a large number of AI support available. This ranges from efficient tools and qualified expert support to broad-based training. For example, the developers receive all the support they need to tackle new AI projects and continuously expand their AI skills.
The central element for this is the AI academy, in which both experienced AI developers and beginners can learn how to use useful tools, frameworks, hardware resources and libraries.
Also, Intel’s Neural Compute Stick 2, which has been specially designed in the form of a USB data carrier for processing in-depth learning calculations, is also beneficial. Above all, they support cost-effective prototyping of image and pattern recognition in the edge. And then there is Intel’s AI Builders. This is a program for software manufacturers (ISVs), system integrators (SIs), OEMs, and companies looking to accelerate AI’s adoption on Intel platforms.
