Introduction

: Data Mining is increasingly popular in the health field because there is a need for an efficient analytical methodology to detect unknown and valuable information of health data.

Objective

: To develop a predictive model using data mining techniques, specifically Decision Trees, to investigate patients with a propensity to develop Type II Diabetes, Arterial Hypertension or Dyslipidemia.

The data of adult patients presenting Type II diabetes, Hypertension or Dyslipidemia being followed in a preventive cardiovascular control program were analyzed with the aim of unveiling phenomena that could help develop the prediction of these risk factors.

Results

: With respect to other decision tree algorithms, Algorithm C 5, showed a greater predictive power. The variables age and waist circumference had the greatest power of discrimination for DM2, HTA or DLP. The C 5 algorithm reached a global precision of 83.01% in the test partition. Then, in the same partition the model managed to discriminate a patient with some of the risk factors in 85.25% of cases, and to rule out any of them in 80.27% of cases.

Conclusion

: Data Mining, specifically decisión tree models, is a valid alternative for early detection of cardiovascular of risk factors.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )