Biomedical Data Analyses Can Predict Metabolic Risk
Analyses of biomedical data from nearly 37,000 volunteer employees of a large company insured under Aetna shows a success rate of 80 percent to 88 percent in predicting risk of metabolic syndrome, which can cause chronic disease.
Analyses of biomedical data from nearly 37,000 volunteer employees of a large company insured under Aetna shows a success rate of 80 percent to 88 percent in predicting risk of metabolic syndrome, which can cause chronic disease.
Metabolic syndrome means an individual has at least three of five biological characteristics that are out of normal range--waist circumference, blood pressure, elevated triglycerides, low high-density lipoproteins and increased insulin resistance--according to a report on the findings in The American Journal of Managed Care. Research published in 2006 suggests that almost a third of U.S. adults have three of the out-of-range characteristics and another 45 percent have one or two risk factors, the study notes. Individuals with metabolic syndrome are twice as likely to develop cardiovascular disease and five times more likely to get diabetes mellitus.
Aetna for several years has offered metabolic screening to covered members, says Adam Scott, managing director at Aetna Innovation Labs. For the study, the labs teamed with data analytics vendor GNS Healthcare to assess the feasibility of predicting metabolic syndrome. Data came from metabolic screenings, insurance eligibility records, medical and pharmacy claims, lab results, and answers to health risk assessment questionnaires.
In the study, claims data proved to be approximately 80 percent accurate in predicting metabolic syndrome and claims data combined with biomedical data had an 88 percent accuracy rate. Researchers also analyzed several scenarios to find if there were any cause-and-effect relationships, says Colin Hill, CEO at GNS Healthcare.
We demonstrated that improving waist circumference and blood glucose yielded the largest benefits on subsequent risk and medical costs, study authors explained. We also showed that adherence to prescribed medications, and, particularly, adherence to routine scheduled outpatient doctor visits, reduced subsequent risk.
Study authors also noted that application of advanced analytics to large data sets is becoming more common in healthcare as the volume and variety of data expand while analysis-related costs drop. The technology and comprehensive data is enabling researchers to generate insights faster and get results to healthcare organizations quickly. The insights derived in the study outlined here were reached in 3 months (1.5 months per model), as opposed to the years that clinical trial and longitudinal studies take.
The study, Novel Predictive Models for Metabolic Syndrome Risk: A Big Data Analytic Approach, is available here.
Metabolic syndrome means an individual has at least three of five biological characteristics that are out of normal range--waist circumference, blood pressure, elevated triglycerides, low high-density lipoproteins and increased insulin resistance--according to a report on the findings in The American Journal of Managed Care. Research published in 2006 suggests that almost a third of U.S. adults have three of the out-of-range characteristics and another 45 percent have one or two risk factors, the study notes. Individuals with metabolic syndrome are twice as likely to develop cardiovascular disease and five times more likely to get diabetes mellitus.
Aetna for several years has offered metabolic screening to covered members, says Adam Scott, managing director at Aetna Innovation Labs. For the study, the labs teamed with data analytics vendor GNS Healthcare to assess the feasibility of predicting metabolic syndrome. Data came from metabolic screenings, insurance eligibility records, medical and pharmacy claims, lab results, and answers to health risk assessment questionnaires.
In the study, claims data proved to be approximately 80 percent accurate in predicting metabolic syndrome and claims data combined with biomedical data had an 88 percent accuracy rate. Researchers also analyzed several scenarios to find if there were any cause-and-effect relationships, says Colin Hill, CEO at GNS Healthcare.
We demonstrated that improving waist circumference and blood glucose yielded the largest benefits on subsequent risk and medical costs, study authors explained. We also showed that adherence to prescribed medications, and, particularly, adherence to routine scheduled outpatient doctor visits, reduced subsequent risk.
Study authors also noted that application of advanced analytics to large data sets is becoming more common in healthcare as the volume and variety of data expand while analysis-related costs drop. The technology and comprehensive data is enabling researchers to generate insights faster and get results to healthcare organizations quickly. The insights derived in the study outlined here were reached in 3 months (1.5 months per model), as opposed to the years that clinical trial and longitudinal studies take.
The study, Novel Predictive Models for Metabolic Syndrome Risk: A Big Data Analytic Approach, is available here.
More for you
Loading data for hdm_tax_topic #better-outcomes...