A new collaboration known as the Pittsburgh Health Data Alliance announced this week a mission of leveraging data to revolutionize healthcare. In the meantime, it is certain to revolutionize workflow at the participating institutions.
The University of Pittsburgh, Carnegie Mellon University and UPMC will be drawing on researchers, physicians and healthcare IT personnel in their collaborative efforts.
The alliance will initially include two research and development centers: the Center for Machine Learning and Health (CMLH), led by founding director Eric Xing, PhD, a CMU professor in the Department of Machine Learning; and the Center for Commercial Applications of Healthcare Data (CCA), spearheaded by Michael Becich, MD, PhD, chair of the Department of Biomedical Informatics at Pitt.
Xing spoke to Healthcare Dive about some of the initial plans, in which his team seeks to revolutionize the concept of basic healthcare data storage to include analysis of the data to provide patterns and predictions, make connections between patients and physicians, etc. The system will use artificial intelligence to generate knowledge and reasoning from the data.
What to do with all that data
Xing says what is unique about this effort is how the data will be processed and used in the long run.
For example, he says, when a patient diagnosed with a particular type of cancer goes to a doctor, the typical experience is that he is seen as one of a group of people who have that same type of cancer, and treated accordingly.
However, this intelligent database would also consider information about the patient which could include data collected through mobile devices to provide insights into the patient’s lifestyle, dietary habits, demographics of their home region, genetic history, and so on.
"That creates what we call a rich and high-dimensional description of you," Xing says, "and that creates the opportunity to look at each patient as a unique individual instead of just one of a group of people." The data, used through an artificial intelligence system, allows for a personalized evaluation of each patient's diagnosis, risks and optimal course of treatment.
At the same time, Xing notes, "This particular individual who is unique is not only to be analyzed as a single instance; he will be connected to millions of patients in the database in a sophisticated way because every individual is similar to some other individual in a particular dimension, such as symptoms or lifestyle, so the patient can be viewed both as an individual and viewed in the context of people with similarities."
Xing suggests that it will be akin to the patient experiencing not only one doctor and that doctor's limited experience with similar patients, but experiencing an interaction with thousands of doctors and millions of patients, so he can be studied in a personalized way within that bigger context.
First projects
Xing says the alliance has been making long-term plans for the coming few years, as well as immediate plans so they can roll out products and prototypes every year.
One thing they're working to do immediately is to push out a doctor/patient question answering system.
The computer system will be able to provide an assessment of the nature of the disease that the patient may have through the patient's answers, and the doctor will be able to get additional questions and recommendations from the computer to help nail down even further the nature of the disease.
"That's the system we're trying to build in the first phase of this project," Xing says.
He says the group is also looking at developing mobile phone devices or other wearable devices that could help the healthcare monitoring system collect patient data such as biometrics and lifestyle information—perhaps including their purchases and where they go, for example—to help gather a holistic picture of the patient's condition.
In order to proceed, they are looking at involving patients from UPMC hospitals, though Xing says they still have to work out the logistics of how to protect the privacy of the patients and how to make the gathering process less intrusive.
Xing expects UPMC hospital IT personnel to be integral to the alliance's efforts. "Their IT people will of course be involved in every phase of the development because it is only through them such products can be deployed or can be experimented," Xing says.
As a final note, Xing looks to address any concerns that this type of work aims to replace much of the human element in healthcare provision by substituting people with machines.
"We are using the system to bring patients and doctors closer through this vehicle of intelligent data analytics," he says. "I don't want to create an illusion or misunderstanding that we're trying to remove the doctor or other human being from this process. In fact, they are more engaged by using such a system."