Fig. 2: Temporally stratified low-dimensional embedding of patients’ diagnosis profiles reveals some separation based on patients’ male infertility status. | Communications Medicine

Fig. 2: Temporally stratified low-dimensional embedding of patients’ diagnosis profiles reveals some separation based on patients’ male infertility status.

From: Leveraging electronic health records from two hospital systems identifies male infertility-associated comorbidities across time

Fig. 2

a UMAP of patients’ diagnosis profiles for diagnoses first obtained before the 6-month cutoff at UC, colored by male infertility status (n = 14,812 patients; 1614 diagnoses as features). b UMAP of patients’ diagnosis profiles for diagnoses first obtained before the 6-month cutoff at Stanford, colored by male infertility status (n = 7723 patients; 1493 diagnoses as features). c UMAP of patients’ diagnosis profiles for diagnoses first obtained after the 6-month cutoff at UC, colored by male infertility status (n = 9234 patients; 1562 diagnoses as features). d UMAP of patients’ diagnosis profiles for diagnoses first obtained after the 6-month cutoff at Stanford, colored by male infertility status (n = 5564 patients; 1452 diagnoses as features). For each panel, the bottom violin plot shows distribution of UMAP component 1 based on male infertility status; the right violin plot shows distribution of UMAP component 2 based on male infertility status. Mann–Whitney U tests assessed whether UMAP components significantly differed based on male infertility status, with a significance threshold of p-value < 0.05. Phenotypes correspond to diagnoses. Pink dots—patients with male infertility, blue dots—patients with a vasectomy-related record, UMAP —Uniform Manifold Approximation and Projection, UC University of California.

Back to article page