Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

Big Data Analytics ID People Risking Metabolic Syndrome

Calculator keys (


27 June 2014. Researchers from the insurance provider Aetna Inc. and GNS Healthcare, a data analytics company in the health care industry, developed statistical models that can identify population groups and individuals at risk for metabolic syndrome, a collection of conditions pointing to future heart disorders and diabetes. The team from Aetna’s Innovation Labs in Hartford, Connecticut and GNS Healthcare in Cambridge, Massachusetts published its findings yesterday online in the American Journal of Managed Care.

Metabolic syndrome is the name given to five related conditions which, when taken together, substantially raise the risk of an individual developing heart disease, stroke, or diabetes. The factors are a large waistline from excess fat in the stomach area, high triglyceride levels, low high-density lipoprotein (HDL or “good”) cholesterol, high blood pressure, and high fasting blood sugar considered an early indicator of diabetes.

Any one of the five metabolic syndrome factors increases the risk of heart disease, stroke, or diabetes, but individuals with more than one of these conditions are considered at higher risk, with the risk increasing as the number of factors increases. The authors cite recent studies showing one-third of U.S. adults — about 80 million people — meet the criteria for metabolic syndrome, with an additional 45 percent with 1 or 2 risk factors for metabolic syndrome.

The researchers analyzed some 37,000 medical records of workers from employers with Aetna health plans who voluntarily took part in metabolic syndrome screening programs in 2011 and 2012. Data collected include insurance eligibility records, medical claims, pharmacy claims, demographics, and lab tests, as well as metabolic syndrome screening results and health risk assessment responses.

With this data set, the researchers applied GNS Healthcare’s analytical platform it calls reverse engineering and forward simulation or Refs. GNS Healthcare says the reverse engineering part of Refs extracts network models from large sets of data with an algorithm that samples successively from a target population. Each sample taken is dependent on the previous sample, thus enabling the algorithm to learn more about the population with each subsequent sample. The company uses a variation of this technique, called the Metropolis algorithm (named after its inventor, not a location), that allows Refs to take random samples from complex distributions, such as those found in health care.

The forward simulation part of Refs tests these models for causation by asking questions of the data about specific interventions. The simulations return probabilities of success, thus the name Monte Carlo given by statisticians to this simulation technique. GNS Healthcare says its use of Monte Carlo simulations makes it possible to return confidence intervals — i.e., margins of error — for each probability.

The analysis focused on two major queries for the data set:

– Predict the probability of each of the five metabolic syndrome variables occurring in individual subjects, based only on claims data.

– Indicate the likelihood of individual subjects improving, staying the same, or getting worse on each metabolic syndrome variable.

The researchers report being able to construct detail risk profiles for the population as a whole and individual subjects, with the calculation of a percentage probability for members developing the five metabolic syndrome risk factors. Calculations include probabilities of developing specific risk factors in the future, based on an individual’s current risk profile.

The models developed in the study also made it possible to test possible interventions to reduce an individual’s risk of developing metabolic syndrome. The researchers found reducing waist size and blood glucose levels return the largest benefits in lowering risks of metabolic syndrome and cutting medical costs. In addition, scheduling a visit to a primary care physician lowers the probability of developing metabolic syndrome in the next year by almost 90 percent.

Read more:

*     *     *

Comments are closed.