top of page



The use of automated exploratory data analysis helped us to cope with a large and complex list of clinical and microbiological variables ...

Dr. Cristina Prat, MD PhD, Head of Respiratory Tract Infection and Mycobacteria (Hospital Universitari Germans Trias i Pujol)

What Is This Work About?


Dr. Alicia Lacoma is a post-doctoral CIBERES researcher and Dr. Cristina Prat is Head of Section of Respiratory tract infection and Mycobacteria at Hospital Universitari Germans Trias i Pujol (Badalona, Spain).

One of the fundamental objectives of the research group is the study of host-pathogen interactions during lower respiratory tract infections and tuberculosis in order to find novel targets for improving diagnosis and treatment. 

The main objective of this study was to evaluate the potential relationships between clinical/epidemiological data and genetical microbial features in a cohort of patients under mechanical ventilation and isolation of Staphylococcus aureus in the respiratory tract .

Dr. Cristina Prat

Dr. Alicia Lacoma

Association between the consecutive isolation of Pseudomonea Aeruginosa with a significantly larger number of days of persistence

What Problem Did We Face?


We combined up to 136 variables related to demographics, clinical evolution, severity scores, biomarkers, follow-up and DNA microbial array from 148 patients. 


Given the exhaustive exploratory approach of this study, the complex list of clinical and microbiological variables and the large number of possible patient stratification factors, we decided to apply and automated EDA tool like AutoDiscovery.

Which Was The Contribution Of AutoDiscovery To This Work?


AutoDiscovery helped us unveiling a clinically relevant association between the consecutive isolation of Pseudomonas aeruginosa with a significantly larger number of days of persistence. Other associations found involved the presence of specific bacterial genes and clinical variables. Some of these associations were only reported in specific patients subgroups.


In this work, we first employed a classical statistical package selecting specific variables collected, and later performed an extensive automated exploratory data analysis combining all host and pathogen data collected.


The key features of AutoDiscovery applied to this work were:


  • The non-guided exploratory strategy proposed by the software, which has proved its effectiveness to analyze and compare multiple variables starting with a set of exploratory questions.





Would you like to contact Alicia and Cristina?


IGTP Campus Can Ruti Ctra de Can Ruti 

Camí de les Escoles s/n

08916 Badalona (Barcelona, Spain)


bottom of page