Page 65 - AIH-2-4

Artificial Intelligence in Health Synthetic data for obesity level prediction

Figure 14. Distribution of obesity level

Figure 15. Age distribution

Figure 16. Height distribution

As illustrated in Figure 21, the axes of the graph and weight (as shown in Equation I and Figure 21), the
represent height and weight, with color coding used to classes appeared to be linearly separable in the graph.
indicate different obesity levels. It is evident that, for a However, including these attributes in model training may
given height, the obesity class increased with weight, and lead to an overestimation of performance. To illustrate this
male subjects tended to fall into higher obesity categories potential discrepancy, two distinct datasets were used: one
compared to female subjects. As BMI – the metric used that includes height and weight, and another that excludes
to assign class labels – is directly correlated with height them.

Volume 2 Issue 4 (2025) 59 doi: 10.36922/AIH025140027