Fig. 4: A parity plot is presented to compare the true values of mPAP (denoted as \({\mathcal{P}}\)) with those predicted by the SISSO model. | npj Digital Medicine

Fig. 4: A parity plot is presented to compare the true values of mPAP (denoted as \({\mathcal{P}}\)) with those predicted by the SISSO model.

From: Artificial intelligence-driven multivariate integration for pulmonary arterial pressure prediction in pulmonary hypertension

Fig. 4: A parity plot is presented to compare the true values of mPAP (denoted as $${\mathcal{P}}$$ P ) with those predicted by the SISSO model.

The shaded region along the diagonal represents a ±5% error margin. Histograms on the top and right margins illustrate the distribution of mPAP across different PH subtypes. The horizontal and vertical gray dashed lines mark the clinical threshold of 20 mmHg for both the predicted and true values, respectively, providing a qualitative visual reference for assessing the incidence of PH. Distinct PH subtypes are denoted using different marker shapes: circles for non-PH, squares for PAH, diamonds for PH associated with lung diseases and/or hypoxia, upward triangles for PH due to pulmonary artery obstructions, pentagons for PH of unclear or multifactorial mechanisms. The performance of the SISSO model is quantitatively assessed using R2, MAPE, and the p value.

Back to article page