A few weeks ago, researchers at Carnegie Mellon University and the University of Pittsburgh announced that they will be teaming up with the University of Pittsburgh Medical Center (UPMC) to tackle big data in healthcare. Through this partnership, called the “Pittsburgh Health Data Alliance,” researchers like Carnegie Mellon’s Eric Xing, PhD, will work to develop new technologies, research and services around big data.
Xing, a professor in the Department of Machine Learning, will lead the initiative at Carnegie Mellon. This week, we sat down with him to talk about the alliance, big data, and his role in this research.
Q: We hear a lot about big data in healthcare these days. Can you tell me about the reasoning behind this alliance in particular? What do you hope to achieve?
The alliance represents an ambitious plan to revolutionize how healthcare can be practiced through the use of big data. Currently, healthcare is practiced in an experience-based fashion. Physicians treat patients based on their own experience and clinical interactions. But, with the knowledge big data provides, we can change the way physicians treat patients. The accuracy of patient care would improve with big data because you can see examples and instances of diseases in other patients, and make recommendations based on those patients. Physicians can also diagnose patients based on patterns that appear in big data that move beyond the individual physician experience.
We are also going to employ artificial intelligence on big data databases to extrapolate the data. For example, since we know every patient is different and is defined by their unique symptoms, behaviors and experiences, we cannot assume one patient’s data will apply to a larger population. If we used that information in a query, we wouldn’t get anything because the patients would be too different. Instead, we’ll use artificial intelligence to make plausible predictions based on the data. We want every patient to receive a highly personalized treatment plan based on statistically high-competence predictions for their treatment and lifestyle.
Q: Why do we need to use big data to manage our healthcare system? How will it help us with patient care?
On the positive side, big data has give us more knowledge. With more knowledge, our position is more informed on patient care issues. For example, in personalized medicine, you can use this data to draw more competent predictions about patient care. On the negative side, people worry about privacy and confidentiality. Big data poses a potential risk because you can disclose confidential information. This is a risk that should not be exaggerated, though – it is more of a practical concern. With good regulation and technology, if the data is protected in a well-designed computer system with good security measures and regulations, the risk of breeching patient privacy can is small. And of course, any benefits always come with risk.
Q: We do hear a lot of criticism surrounding privacy issues. Can you talk more about that? What are the challenges associated with managing big data?
The management of big data is always a complex issue. Once the data becomes, big you need to have large infrastructures for convenient storage and accessibility. But there’s a tradeoff – if you want it to be convenient for the physician to access data via a handheld device, the security for the data becomes more difficult to maintain – the data could be more easily hacked. The real challenge for technologists is to identify loopholes in data storage, connection and availability. Carnegie Mellon scientists are quite aware of these risks and this is one of our major projects in this initiative. We are going to attempt to study these risks and potential vulnerable points using our computer science technology.
Q: What do you think will be the key to successful research in this initiative?
This particular type of research has a very clear goal for serving people’s needs, which is different from traditional academic research. Traditionally, we deliver research by writing papers and reports that are reviewed by peers and then presented to the community. However, we need to interface with stakeholders and the clients for this project. In my current vision, we are trying to string together open channels between the back-end researcher and the front-end consumers so that the client’s needs are communicated. I want people to understand each other and align to work towards a common goal.
Q: Last question. What are your personal goals – in this initiative, and otherwise?
I want to see the success of the initiative and the Center. I would like to see the Center become an incubator for a new kind of science that connects information technology with healthcare and medicine. I would be very happy if this new platform was made available to health providers, health receivers, and researchers. If I can jumpstart this project and set up a reasonable initial agenda to lead the way for the future, that would be ideal.
Soniya Shah is an on-staff contributing writer at MedTech Boston. She's a senior at Carnegie Mellon University pursuing a BS in technical writing. She has experience as a ghost writer and medical writer, and in developing software documentation.
Send this to a friend