Michael Yao, Pulse Writer | December 27, 2017
This past November I had the opportunity to attend Transforming Healthcare with Data, a data science conference hosted in conjunction with the University of Southern California (USC). I was able to hear experienced programmers and healthcare professionals discuss their views about the healthcare field and what the future might entail.
For the past few years, healthcare has been in a transitional state of flux.
We are in a position where traditional ‘pen-and-paper’ methods of storing patient personal information have not been abolished, but modern software-based methods have not been completely implemented. As we inevitably make this transition, we need to consider where digitally-based methods could fail and how we can proactively ensure that large-scale data sets can be effective tools to help solve the medical problems we face in healthcare.
The idea of ‘digital health’ infuses medicine with technology. In addition to the advent of electronic health/medical record (EHR and EMR) software, which streamline the storage and analysis of patient data, digital health means big data and how we can make statistical inferences across entire populations.
Ultimately, digital health relates to the idea of the ‘healthcare ecosystem’—a compilation of public and private healthcare payers, hospital administrators, healthcare providers, pharmacies, and medical research centers, all working to benefit patients and associated consumers.
But the question still stands in the field of digital health of how we can take aggregates of information, interpret them, and then feed our results back to patients for their benefit: the ‘problem of big data.’
Not surprisingly, many industry professionals agree the solution lies with artificial intelligence (AI). According to Terry Sanger [1], “the underlying mechanisms behind machine learning are evolving. It used to be that machines could simply be given a set of rules to follow for data analysis, but this is no longer the case…the amount of personal data is exploding.” Indeed, according to estimates from Arvid Raulinaitis [2], clinical factor data comprises approximately 0.4 terabytes (TB), genomic factor data 6 TB, and exogenous factor data a whopping 1100 TB! We need to change our algorithmic methods for data analysis to account for the exponential increases in the sheer amount of data being generated.
Jerry Power [3] and other experts believe that the answer to our problem of big data may lie with the Internet of Things (IoT), a network of interconnected digital devices that have the unifying capability to communicate with one another.
When we realize that many participants of the healthcare ecosystem interact and share data to optimize patient outcomes, we can more clearly see why the Internet of Things can be such a powerful tool. We need to be able to track the inputs of doctors, nurse practitioners, families, emergency service providers, labs, and pharmacies all in real time.
Power also asserts, “IoT data processing can be applied to known data to uncover important relationships, and is constantly appended by new data from sensors embedded in devices, looking at data and situations as they are occurring.”
There have already been efforts to pursue large-scale IoT implementation. The Intelligent Internet of Things Integrator (I3 for short) is a joint project led by USC’s Marshall School of Business and Viterbi School of Engineering, supported by both the LA Children’s Hospital and City of Los Angeles.
Simply put, I3 is the intersection of public and private sectors—an open-source, community driven development effort that can only be described as an open IoT data marketplace that uses devices to accept data and connects them to many interlinked applications.
I3 promises to reduce central cloud and smartphone application management needs by managing IoT security close to the individual devices without forcing on intricate software requirements. Furthermore, I3 plans to incorporate AI systems extensively, such as to determine consoles or applications that have suspicious activities. Essentially, I3 will transform the Internet from a connectivity tool to a context awareness tool. Check out I3’s web page here.
One of I3’s primary functions will be to organize the many health ecosystems currently present into a singular, coherent system. As a natural consequence, data ownership will be delimited, which naturally brings I3’s security protocols into question. The potential issue ultimately boils down to one of trust.
The founders of I3 believe that they have incorporated this critical component of trust in a unique way. In I3, application owners rate device owners for data fidelity, who in turn rate applications for trustworthiness. Trust is ensured and enforced by the intrinsic transparency of the I3 community.
However, trust is not merely dependent on addressing software concerns. The situation is further complicated by the complex psychosocial and behavioral factors that affect people’s trust of internet-based systems. Everyone’s definition of privacy is different, and no one will simply ‘give up’ their personal information to a nameless cloud in the sky. People need to feel comfortable in an environment where they are willing to share their data of their own accord, a factor that can only be addressed by changing our perceptions of modern technology—both its merits and limitations.
There are also data management challenges with the advent of IoT. With diverse data types coming from different sources, we need a method of data sorting and aggregation that allows for the continuous collection of data from integrated assets. How can we ensure device and data security? Can we preserve the ability to integrate technology yet to be invented? How can poorly equipped legacy systems incorporate IoT and rapidly increasing data volumes and their velocity? How can we efficiently draw meaningful insights from these data volumes?
We need solutions purpose-built to address these challenges and tackle a wide range of IoT data sources—solutions driven by advanced analytics, interoperability, data lifecycle management, and open-source solutions. Although there are many issues we still have to consider, IoT seems to be leading us on the right path.
As best put by Jerry Power, “big data is emerging as a new force in the healthcare industry, and the Internet of Things will drive the big data revolution further and farther.”
1. Terry Sanger is the Academic Director of the Health, Technology, and Engineering graduate program at USC.
2. Arvid Raulinaitis is the Cognitive Architect and Project Manager in IBM’s Cognitive Business Decision Support.
3. Jerry Power is the Executive Director at USC’s Institute for Communication Technology Management.
Any questions or comments can be sent to myao@caltech.edu or medlife@caltech.edu
________________________________________________________________________________
ABOUT THE AUTHOR Michael Yao is a first-year undergraduate student at Caltech, and intends on pursuing a future career in medicine. In addition to writing for the Pulse, Michael is an Associate Editor for the Undergraduate Research Journal, a member of Caltech’s Swim and Dive Team, and a part of the Academics and Research Committee. In his free time he enjoys running, nature hikes, and spending time with his dogs at home.