Age Estimation from Blood Test Results Using a Random Forest Model

Abstract

Abstract Background and Objectives From the perspective of preventive medicine, in situations where screening tests are widely used, this study aims to clarify the role of screening data on ageing and health problems by estimating age from screening data with verifying the number of data items required. Materials and Methods A Python random forest model was generated using Chat GPT and tested. Results When using all 71 items, including gender, for the test results, a high accuracy of R 2 = 0.7010 was obtained when there were 9243 training data sets (80% of the total number of data sets). The R2 decreased slightly to 0.6937 when the number of data items was reduced to 15 by discarding lesser importance items. When the number of data sets were less than 800 or when the number of data items were less than 7, the R 2 value fell below 0.6. Interestingly, a higher age was tended to be estimated for post-menopausal women compared to pre-menopausal women. Conclusions The age estimated from blood data by the random forest model (blood age, so to speak) is so precise that it can be useful for assessing physical ageing state. However, the specific relationship between blood age and health status is still unclear, waiting for future research in order to deepen our understanding of this area.

Ähnliche Arbeiten