[ad_1]
You have undoubtedly filled countless patient information sheets in waiting rooms at hospitals or medical centers.
Usually they ask you for an emergency contact – and for many people, it's a parent.
In most cases, the closest data are never re – examined except in an emergency. But what if you could link people according to their list of blood relatives?
This was the purpose of a recent research article published in the journal Cell
The result was the construction of vast multigenerational networks. According to Nicholas Tatonetti, an assistant professor of biomedical informatics at Columbia University in New York, "we can use an automatic algorithm to find these relationships for hundreds of thousands of people.
"We can now evaluate thousands of different conditions that simply would not have been possible.
And here in Australia, experts hope that the data from My Health Record could be used to a great extent. in the same way – as a powerful tool for medical research.
How does it work?
Dr. Tatonetti's research involved the use of a computer algorithm to scan the records of millions of patients at three years of age. rk hospitals – Columbia, Mount Sinai and Weill Cornell
The analysis found 7.4 million family relationships in more than 3 million patients. Many family trees have had more than 30 people, spanning several generations.
But family ties, although interesting, were not the main game for researchers.
"This means we can track the disease in this family and study how the disease could be inherited," said Dr. Tatonetti.
19659002] "This allows us to study many different traits – essentially everything that is captured when you encounter the health system. Then we can group them into all those families and learn something about how the disease in general is inherited. "
For the most part, the data on family trees matched what had already been established in the medical literature – diseases such as heart disease, diabetes, and obesity were all as hereditary as previous studies suggested. 002] But he also provided clues about conditions that might be more heritable than we thought.
"We were able to do a comparative analysis of the heritability of cholesterol, which had not been fully studied before". 19659002] "We have been able to show quite clearly that HDL [good cholesterol] is more heritable than LDL [bad cholesterol] and this could have implications for the way we treat and manage cholesterol in the future."
Can we do something like that in Australia?
Australia records similar data to those used in the US study.
This is what is called "routinely collected data", and that is the information that is generally collected whenever a patient visits a hospital .
But Louisa Jorm, director of the Data Research Center at the University of New South Wales. said the data is not readily available for medical research.
"We have the same data captured in the patient administration systems of all our major hospitals," said Professor Jorm
. there is a huge amount of work to access these data and put them in the form that would allow this research. "
The deployment of My Health Record could help advance this type of research," said Professor Jorm.
"There are challenges, because there are big problems with data quarrels," she said.
"Much of the data that will be in My Health Record is in the form of PDF documents and texts, the areas in our center that we are trying to build is how do we actually extract the value of these data from type text.
"Knowing that we are going to have a national health record for the entire population, it is a tremendous research resource, potentially, but a lot of work to be done to extract that value." [19659002] The Department of Health has published a framework to guide the use of My Health Record data for secondary purposes, such as research, in May.
What about life? Private Patients – Many health professionals are excited about the opportunities that My Health Record offers for research, concerns have also been raised about the deployment of the opt-out and the health care system. access to data. [19659002] In the US study, data were evaluated in 18 disidentification process steps after being analyzed by the computer algorithm and before the analysis was communicated to the researchers.
This includes the deletion of names, geographic data, medical registration numbers and social security numbers.
"When we analyze the data and establish those connections, it's an algorithm that does it blindly." And then all the data is mapped and depersonalized, so no personal information is there before that. They were not analyzed later. "
However, researchers recognized that it was possible to re-identify data in limited circumstances – for example, if things like birth dates could be cross-checked with birth announcements and death notices to identify unique family trees in the data set.
to re-identify individuals in a supposedly depersonalized dataset published by the federal Department of Health.
Professor Jorm states that the federal government consider a more rigorous approach to protecting health care data, and more generally response to the Productivity Commission's report on the availability and use of 2016 data.
In this system, it's not just the disidentification of the data that matters, but all of it. data processing infrastructure. "It's what's called a five-level security approach, in which you do not just rely on a set of protections, for example, we trust individuals. We rely on the technical controls, the trust, the accreditation of the individuals, and then the data themselves and what we can do to reduce the risk of identifying these data, "he said. declared Professor Jorm
. comprehensive approach to how we protect privacy and a much stronger national framework on how we do it as a result of the government response announced in the last federal budget. "
Dr. Tatonetti believes that it is important" In the future, we will start to see these data come together and get used to patient care, "he said.
Until we can really tackle how to handle all this data, to all the privacy issues and all the challenges of analyzing these big datasets, let's go
"We have many interesting challenges to overcome."
ABC
Source link
