Discovering multiple antibiotic resistance phenotypes using diverse top-k subgroup list discovery

Area of Science:

• Medical Informatics
• Computational Biology
• Public Health

Background:

• Antibiotic resistance is a critical global health challenge, necessitating advanced clinical decision support.
• Current patient phenotyping methods may yield single, potentially insufficient, explanations for complex medical phenomena like antimicrobial resistance.

Purpose of the Study:

• To define and address the problem of mining diverse top-k patient phenotypes for clinical decision support.
• To introduce the EDSLM algorithm for discovering multiple and varied phenotypes associated with specific medical conditions.

Main Methods:

• The study proposes the EDSLM algorithm, integrating Subgroup Discovery, the subgroup list model, and the Minimum Description Length principle.
• The algorithm is designed to mine diverse sets of patient phenotypes from clinical data.

Main Results:

• The EDSLM algorithm enables the discovery of multiple, distinct phenotypes for a given medical phenomenon.
• Demonstrated a practical application of phenotyping in antimicrobial resistance using the MIMIC-III dataset.

Conclusions:

• The developed method provides clinicians with a valuable tool for obtaining diverse patient phenotypes, enhancing the understanding of antimicrobial resistance.
• This approach supports more comprehensive clinical decision-making in the face of evolving antibiotic resistance patterns.