Analyzing Candidemia in Critically Ill Patients: Unsupervised Clustering Reveals Three Distinct Phenotypes
-
By
-
Zejun Yang
-
Yaotian Zhang
-
August 28, 2026
Clinical Scorecard: Analyzing Candidemia in Critically Ill Patients: Unsupervised Clustering Reveals Three Distinct Phenotypes
At a Glance
| Category | Detail |
|---|
| Condition | Candidemia in critically ill patients |
| Key Mechanisms | Unsupervised machine learning identifies distinct clinical phenotypes with varying mortality rates. |
| Target Population | Critically ill patients with candidemia |
| Care Setting | Intensive care unit |
Key Highlights
- Three distinct phenotypes identified: immunosuppressed hematological patients, deteriorating cirrhotic patients, and catheter-related candidemia patients.
- Phenotypes 1 and 2 have approximately 70% mortality, while Phenotype 3 has 50% mortality.
- Methodological weaknesses in the study design raise questions about the clinical utility of the identified phenotypes.
Guideline-Based Recommendations
Diagnosis
- Consider clinical phenotypes when assessing candidemia outcomes.
Management
- Future models should exclude broad clinical scores linked to mortality from baseline features.
Monitoring & Follow-up
- Use validated baseline factors to assess clinical relevance post hoc.
Risks
- High mortality rates in immunosuppressed and cirrhotic patients.
Patient & Prescribing Data
Critically ill patients with candidemia
Targeted immunomodulatory or antifungal therapies may be developed from unbiased clinical profiles.
Clinical Best Practices
- Avoid using composite severity measures like SAPS II in unsupervised clustering for candidemia.
- Utilize unbiased clinical, transcriptomic, or biomarker profiles for future models.
Related Resources & Content
