Prediction of disease - really

American scientists were able to develop a special algorithm that allows physicians to predict the patient's disease, according to EurekAlert! On this algorithm worked Tyler McCormick and his colleagues at the University of Washington. All the results are published in the journal Annals of Applied Statistics.

To develop a statistical model, Mak karmic and his colleagues studied data of many thousands of patients, age category which amounted to forty years. They all participated in the clinical trials of a variety of medical drugs. In each case history, which was taken into consideration, were available data on all diseases, and the drugs that were sold to the patient. In addition, it indicated the demographic data (race, gender, and other).

Based on the theory of probability - Bayes ' theorem - researchers have developed a special algorithm, called HARM (Hierarchical Association Rule Model). This algorithm scans than previously ill person, and yet compares its data with many case histories that have similar symptoms. Next, the algorithm predicts what else will soon be hurting the patient. Also, this algorithm can calculate the approximate time of occurrence of the predicted disease.

Subscribe to new posts: