A new study by PhD student Denis Klimov under the direction of Prof. Yuval Shahar represents a new milestone in the ambition of the Research Center for Medical Information Systems at Ben-Gurion University of the Negev to produce a system for the automatic (or semi-automatic) discovery of new medical knowledge from a large amount of medical data.
A new study by PhD student Denis Klimov under the direction of Prof. Yuval Shahar represents a new milestone in the ambition of the Research Center for Medical Information Systems at Ben-Gurion University of the Negev to produce a system for the automatic (or semi-automatic) discovery of new medical knowledge from a large amount of medical data.
Klimov's research was recently published as an article in the leading journal JAMIA, together with another master's student, Alexander Shakanevsky, who also conducted his research under the guidance of Prof. Shahar, from the Department of Information Systems Engineering (Head of the Medical Information Systems Research Center). The article demonstrates that existing medical knowledge can be used, by using interactive data mining technologies and advanced computational technologies from the field of machine learning, to discover new and valuable clinical knowledge.
As part of the article, Klimov and Skanevsky present the use of the VITA-LAB system, which was developed by Klimov as part of his doctoral thesis, at the Center for Medical Information Systems at Ben-Gurion University of the Negev. The system is a combination of two previous systems that were also developed at the center: the VISITORS system (developed by Klimov in his master's degree research), which allows interactive and visual investigation of time-dependent data of a patient population, and the KARMALEGO system (developed by Dr. Robert Moskovitz ' during his doctoral thesis under the guidance of Prof. Shahar) to discover frequent time-dependent patterns that repeat themselves in patient data accumulated over time.
The researchers showed that the integrated system, VITA-LAB, allowed the researchers to discover different time-dependent patterns in the data of about 22,000 type 2 diabetes patients, in addition to the variety of options it supports. It turns out that some of these patterns, which can be discovered and studied with the help of the innovative system, have a predictive value regarding the patients' risk of developing deterioration in their renal function within five years.
Kidney damage, which manifests itself in increased protein excretion in the urine, is one of the most difficult complications among diabetics. It can lead to end-stage renal disease, ESRD. The treatment of the disease is considered one of the main and difficult complications for diabetes patients, and is considered very expensive - for example, in the USA, the treatment of ESRD already cost at least $2010 per year per patient in the government health system "Medicare" in 66,000, and the cost has climbed by about 8% every year since then ( According to the data of the statistical yearbook, USRDS, in 2012).
These days, the researchers have begun to apply the new system to a variety of important medical problems in other clinical areas, such as classifying and predicting infections in patients hospitalized in intensive care, analyzing data from oncology patients, and more.
The VITA-LAB system is part of the analysis and decision support tools offered by the "Medilogos" company established by the AVG application company in collaboration with the Research Center for Medical Information Systems, and is expected to enter clinical use during the coming year.
