Prognostic Phenotype Clustering of Uterine Sarcoma Patients

Uterine sarcoma is a rare but highly aggressive malignancy with poor prognosis. In this study, prognostic pheno-types corresponding to the end point of mortality were explored by unsupervised machine learning (ML) on training <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathrm{n}=1985)$</tex> and validation <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathrm{n}=498)$</tex> datasets. The data were extracted from patients with uterine sarcoma in the Surveillance, Epidemiology, and End Results (SEER) database. A decision-making process of phenotype identification was performed and the findings were compared with existing research. Three phenotypes associated with stratified levels of mortality risks (high, intermediate and low) were identified. Kaplan-Meier survival analysis revealed significant differences among these phenotypes (pairwise comparison <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$p &lt; 0.05$</tex>). The hazard ratio for mortality was 0.77 (95 % CI 0.68-0.88, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$p &lt; 0.001)$</tex> between Phenotype 1 (high risk, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{n}=673$</tex>) and Phenotype 3 (intermediate risk, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{n}=631$</tex>), and 2.8 (95% CI 2.4-3.3, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$p &lt; 0.001)$</tex> between Phenotype 2 (low risk, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{n}=681$</tex>) and Phenotype 3. Furthermore, pathological grade, tumor size, and lymphatic node metastasis were identified as three critical predictive factors by comparing the distribution of risk factors among phenotypes. Without any priorities, ML effectively explored phenotypes that behave differently from data, offering a fresh viewpoint on prognostic prediction in uterine sarcoma.