The information system of MDClone was first installed at Haifa's Rambam Health Care Campus in Israel in the company's first pilot project. CEO Ziv Ofek, CTO Luz Erez, and CFO Boaz Gur-Lavie founded MDClone. Ofek was a cofounder of dbMotion, a groundbreaking digital medicine company sold to US company AllScripts for $235 million in 2013. He also chairs the Center for Digital Innovation (CDI) in Beer Sheva, where MDClone is also located, and Day Two. The system enables researchers and doctors to conduct research using big data databases of patients in hospitals without jeopardizing their privacy.
Ostensibly, it is enough to remove the name and ID number from the information, but this is not the case. Identity specialists, some acting from negative and illegal motives, are able to use several particulars to glean the identity of an individual patient by cross-referencing with other open or leaked databases. The combination of the date of birth, zip code, and gender makes it possible to identify most people in the US, for example.
What MDClone does is to invent non-existent patients with false characteristics and conduct research on them. How can research be conducted on non-existent patients? MDClone chief science officer, former IDF general headquarters Center for Systems Analysis head, and former Israel Air Force Operations Research unit head Hovav Dror explains, "Using sophisticated mathematical models, we create fictitious people for whom the relation between the data that interests us is the same as in the original data, but other parameters are different, so that the particulars on the original record can never be reproduced."
How do you know which data in each record are important to preserve? That is what we want to research - the connection between things.
Dror: "The system scans all the variables and the connections between them, and we preserve all the important correlations. If there are a million variables, we won't do so well, but we manage to preserve real information about the complex connections between hundreds of variables."
When you present the information to the researcher, you have actually already done the work. You have examined the connections and decided where there are correlations, and have brought the researcher what purports to be complete medical records. But the only correlations for which the real information was actually preserved are those that the system has classified as interesting.
"That is true to a large extent. We give them a very widespread network of connections, but only the system knows all the connections, because they are so many and widespread. A researcher is actually asking the system a research question about a specific connection that interests him, while our aim is that only researchers with a high level of access to clinical information can investigate these data, but the information will be very accessible."
Rambam endocrinologist Dr. Irit Hochberg is the first one to conduct research using the new technology, after having conducted a preliminary study to validate the technology's accuracy, based on data from her department. The validation study question involved identification of factors increasing the risks of developing hypoglycemia among hospitalized diabetes patients. Hochberg compared the patients' real data with the parameters found to be important and predictive using the new system, and showed that the results obtained were the same.
Rambam director Professor Rafael Beyar said, "This technology enables us to be in the forefront of science and development quickly and efficiently, while every doctor at the hospital who has obtained authorization can lead research, while preserving absolute medical confidentiality. This will facilitate a breakthrough in medical knowledge, increase the speed and availability of research capabilities, cut costs, and especially completely ensure patients' anonymity."
Hochberg added, "A doctor encounters medical scientific questions on a regular basis when treating patients. MDClone's technology makes research capability directly accessible to the doctor. It shortens the time needed from the idea for a research question until retrieval of the data for answering the question from months to a few hours. In this way, scientific answer can be obtained quickly, while preserving the complete confidentiality of the patients' medical information." Research on people, even if the research is big data, still requires approval from the Ministry of Health.
Published by Globes [online], Israel Business News - www.globes-online.com - on June 5, 2017
© Copyright of Globes Publisher Itonut (1983) Ltd. 2017