Healthcare is one industry where speedier data processing is becoming increasingly vital. Medical practitioners typically have access to a mountain of data, but access to that data (such as clinical records, sensor data and genomic analysis) is often slow and inefficient.
Of course, the arrival of faster storage technologies aims to provide diagnoses and actionable insight faster than ever before. The benefits are twofold. First, patients will have to wait less time for test results, improving the service they receive. Second, doctors can provide more in-depth analysis faster, allowing them to spot issues and start treatments quicker. With more targeted healthcare plans, data analysis can help save patients and provide the care they need.
To see the power of computing in the medical space, you only need to look at MIT’s new algorithm for comparing Magnetic Resonance Imaging (MRI) scans. Comparing two images could typically take up to two hours “as traditional systems meticulously align each of potentially a million pixels in the combined scans” . With MIT’s new algorithm, however, the same comparison can be done over 1,000 times faster.
It’s still a lot of information to crunch and so future advancements in healthcare and precision medicine hinge on faster data processing. Cloudera provides centralised cloud technology solutions for enterprises and, as Chris Darvill, Senior Director for Sales Engineering explains in the video below: “Being able to bring together clinical records, sensor data, patient records [and] genomic analysis... can really drive a better outcome for patients.”
Today, a lot of delay in data processing is down to the speed of storage, with processors often sitting idle while they wait for data to be retrieved from existing HDDs and SSDs. Increasing the speed of storage, therefore, puts data right where it’s needed, when it’s needed, dramatically increasing overall access times and enabling speedier responses.
The power of such low-latency storage can be seen at the University of Pisa. Moving to use Intel® Optane™ technology, the university has managed to dramatically reduce MRI screening times, making the process more comfortable for patients while delivering more accurate results for doctors and technicians. 
Intel® Optane™ DC Persistent Memory allows developers to simplify their database deployment by keeping the entire dataset in a single multi-model database. 
While existing technologies certainly suffice, reading and writing data from traditional HDDs suffers from high latencies. Serial ATA and NVM Express solutions are quicker, but they could not operate fast enough due to the demands of the university’s MRI ‘fingerprinting’ techniques. This involves a process called undersampling, where an MRI scanner outputs less data and fills in the gaps using pre-computed tables. It’s an effective solution if you have an abundance of memory.
But adding large amounts of DRAM is costly. Transitioning to the low-latency Intel® Optane™ SSD DC P4800X Series, the University could extend its computing environment by allowing programs to access files as if they were in memory. As a result, laborious MRI screening times dropped from 40 minutes to just two minutes, while post-scan processing now takes hours rather than days.
The work by the University of Pisa is indicative of a larger shift towards in-memory computing, which uses DRAM to give fast, low-latency access to data. But there’s a catch... Any data held in memory is not persistent and is wiped on a system restart or in the event of a power failure. Combined with relatively low storage volumes and comparatively high DRAM prices, in-memory computing has several restrictions.
For those reasons, 2019 will see a push towards persistent memory (PMEM) technologies, such as Intel® Optane™ DC Persistent Memory. This puts high-capacity storage close to the processor, giving memory-like access speeds, with SSD-like storage retention.
As Redis Labs, a leading supplier of in-memory database technology to top healthcare companies, explains: “Redis Enterprise’s multi-model capabilities, which encompass core data-structures (including the new Redis Streams) and modules (ReJSON, RedisGraph, and RediSearch) all perform at a sub-millisecond response time for any operation with persistent memory.” 
Real-time analysis is becoming a crucial tool in healthcare. For example, Danish startup Corti uses machine learning to analyse emergency calls for signs of cardiac arrest. “We have built an AI capable of listening to and learning from emergency calls,” says the company. “It instantly matches live audio with thousands of past calls, supplementing emergency dispatchers with superhuman pattern recognition.”
Corti looks at the words used by callers, the tone of their voice and any background noises. As a result, in tests across 161,000 emergency calls in Copenhagen, the software could detect cardiac arrests in 93 percent of cases, versus 73 percent for a human operator. 
It’s just one use-case for real-time healthcare diagnostics, but it points towards a future where our healthcare will be largely data-driven. As GridGain explains , there’s a shift towards precision medicine and personalisation, with vast amounts of data required to support these. In particular, there’s more patient-specific data available through health wearables, home monitors and smartphones.
Handling all these data streams to give an accurate diagnosis and a personalised response requires large amounts of fast data, with in-memory databases providing the speed that real-time applications require.
Modern healthcare will increasingly rely on this range of data sources, requiring low-latency, high-capacity data storage. From fast flash arrays to persistent memory, healthcare providers now have access to the huge amounts of storage and processing power required to provide faster diagnostics and more effective treatment plans.
 &  https://devops.com/redis-labs-delivers-the-fastest-multi-model-database-on-intel-optane-dc-persistent-memory/