Nov 01 2018
While BLAST represents a great bioscience use case for computational storage, it is by far not the only great bioscience use case. There are a variety of bioscience problems that are “parallel in nature” where massive data movement is a significant problem, and for which Computational Storage could have value. Radiomics, a field of medical study whose goal is to extract useful features from large numbers of medical images to improve diagnostics, is a perfect example of this. The data sets that are used in radiomics studies can easily be in the petabyte range – for instance, Memorial Sloan Kettering Cancer Center has over 25 million glass pathology slides which are being scanned into their digital database, which are then analyzed to create machine learning models to better diagnose cancer. This is a problem set where computational storage can accelerate performance.
The Internet of Things (IoT) also extends to biology in the form of wearable technologies, forming an “Internet of Beings” (IoB). As massive amounts of data such as heart rate, respiration rates, blood pressure, body temperature, and other metrics are gathered, they can also generate very large amounts of data. Drug discovery processes are also an area where wearable technologies are also expected to produce large amounts of data. For many use cases that desire to utilize data from the “IoB”, the processing of this data must be both fairly rapid and economical, both things that computational storage can help achieve.
To be sure, the correct management of data sets of this size is also an issue which has been covered. Issues such as the elimination of patient identifying information (PII), and the correlation of unstructured data sets such as imagery with clinical notes over time, are clearly one of the issues that must be effectively dealt with in any bioscience petabyte-scale problem, but these are issues for which there are a number of solutions already. The movement of radiomic petabyte-scale data sets between storage and compute however is one that could be uniquely solved by Computational Storage. Find out more on how products from NGD Systems can help solve your petabyte-scale issues on our website.