The Reconfigurable Future of Healthcare
Data plays a more central role in healthcare than ever before. It won’t be long before every person’s genome is sequenced at birth, with follow-up sequencing done at regular intervals throughout life. Each genomic check-up would produce roughly 180 gigabytes of data that will need to be processed, analyzed, and stored. The promise of using such data is that doctors would be able to intercept diseases before symptoms develop, taking preventative medicine to a whole new level.
Already, whole-genome sequencing is helping to diagnose otherwise-hidden diseases, while machine-learning tools can enhance a doctor’s decision-making by scouring reams of data with speed and accuracy that far surpasses human capabilities. Meanwhile, the advent of wearable sensors is generating an unprecedented amount of additional data, including heart rate, respiration, blood pressure, and other vital signs. No wonder we stand at the threshold of healthcare’s “era of data.”
Big data is at the center of this whirlwind, but putting all this data to good use has been another matter altogether. What we have been missing, until recently, has been ample computing power.
In order to handle the surge of useful health care data—genomic and otherwise— available to us, we must adopt more powerful computing tools than the CPU-based machines that have dominated the computer and server industries for decades. Unless we do so, we will continue to face a data bottleneck that stands in the way of quick answers for healthcare providers and researchers.
Fortunately, an answer exists in a technology that has actually been available for decades —FPGAs, or field-programmable gate arrays—which are now going mainstream. Why now? In the past, programming these chips required significant expertise. Now, however, computing demands have reached a point where the efficiency of an FPGA outweighs programming hurdles. The increased use of FPGAs has in turn lowered costs, further accelerating mass adoption.
How do FPGAs achieve such efficiency? It comes down to design. Instead of the many lines of software code that are executed on a CPU, a logic circuit is utilized. This logic circuit can then be replicated many thousands of times, creating a massively parallel computing architecture rather than the minimally parallel nature of a CPU that has only a few cores or threads available. The result is rapid speed, with an output provided nearly instantaneously after an input is applied.
This efficient processing chip is already picking up where Moore’s Law left off, vastly increasing computing speed at a lower price, and providing value across industries. Intel expects the FPGA market to double in the next decade, and is forecasting revenues of $8.9 billion in 2023. FPGAs are now even powering the cloud with Amazon Web Services’ launch of EC2 F1 instances with field programmable gate arrays. It’s been estimated that up to one-third of cloud providers will be using hybrid FPGA server nodes by 2020. FPGAs are already commonly used in aerospace, defense, automobiles, and consumer electronics.
The use of FPGAs in healthcare is still nascent, but in areas where they are being applied, we are beginning to get a taste of their potential to change how medicine is practiced and health is managed. FPGAs provide the ability to analyze a whole human genome with a single computer—down from 80 computers using CPUs—and in a fraction of the time—22 minutes instead of more than 30 hours. FPGAs also enable real-time video during robot-assisted surgeries, meaning doctors obtain instant feedback and do not have to take their eyes off the procedure at hand, increasing safety and number of procedures that can be performed. The benefits of these processors are only starting to present themselves.
At Edico Genome, we created a specialized FPGA-based processor called DRAGEN for analyzing next-generation sequencing data. This chip powers a data analysis platform that is being used to speed turnaround time of genomic tests, resulting in faster diagnoses of critically ill newborns, cancer patients, and expecting parents awaiting prenatal tests. Faster answers also benefit researchers: Scientists at Baylor College of Medicine studying 3-D structures of DNA were able to accelerate by nearly 20-fold the analysis of the massive data sets generated.
One important note for entrepreneurs entering this space is the importance of patent protection. The potential uses for FPGAs are seemingly boundless: Over the past five years, Edico has grown its FPGA intellectual property portfolio to nine issued patents that broadly cover genomics, and an additional 16 are filed and pending with the United States Patent and Trademark Office. Having a strong patent portfolio makes it easier to raise financing from investors, establishes your company as a leader in the industry, and, importantly, deters would-be competitors from following your lead.
Beyond genomics, which is currently the most obvious (and high-demand) application for FPGA technology, FPGAs can become a valuable tool for medical imaging and artificial intelligence (AI) tools for healthcare. AI, in particular, holds great potential to mine big data to improve health care decision-making, and thus will become even more critical in the future. As we are able to produce large quantities of data, we also need help to make sense of it all, and to do so very quickly. AI can do that, but not using sluggish CPUs. After all, healthcare data is most valuable when it can be interpreted and put to use right away.
The potential applications of FPGAs are numerous, and some near-future applications include:
—AI “droid” doctors—imagine pulling up to your grocery store, or better, pulling out a device at home, and receiving a checkup by giving a painless blood sample via pin prick, having it analyzed, compared to prior checkups by scanning your electronic medical record (EMR) and being offered a customized health plan, all in a matter of minutes. The robot doctor could recognize the clues of disease earlier, and recommend interventions before it starts, leading to healthier populations and saving health care costs.
—Automated lab testing — diagnostic testing, including analysis of X-rays, CT scan,s and biopsies, could be automated, improving accuracy and turnaround time while reducing costs and human error.
—Improved mobility — Robotic exoskeletons could become mainstream as speed is improved and costs reduced, which would be life-changing for people using wheelchairs. These exoskeletons currently cost between $40,000 and $70,000, and reach a speed of one mile per hour, which is still below the average walking speed of a human (three miles per hour).
To get to this future though, we’ll need to determine how to digest the data we’re generating and glean impactful insights. I believe FPGAs are the key that will unlock this future and take us forward.