Predicting ICU Admission in Pregnant Women with COVID-19 Using Machine Learning
Author Information
Author(s): Mukhamediya Azamat, Arupzhanov Iliyar, Zollanvari Amin, Zhumambayeva Saule, Nadyrov Kamalzhan, Khamidullina Zaituna, Tazhibayeva Karina, Myrzabekova Aigul, Jaxalykova Kulyash K., Terzic Milan, Bapayeva Gauri, Kulbayeva Saltanat, Abuova Gulzhan Narkenovna, Erezhepov Baktigali Aubayevich, Sarbalina Asselzhan, Sipenova Aigerim, Mukhtarova Kymbat, Ghahramany Ghazal, Sarria-Santamera Antonio, Cucinella Gaspare
Primary Institution: Nazarbayev University
Hypothesis
Can machine learning techniques predict whether a COVID-19-infected pregnant woman will be admitted to ICU?
Conclusion
The predictive model may efficiently support the prioritization of care for COVID-19-infected pregnant women in clinical practice.
Supporting Evidence
- 10.4% of the analyzed pregnant women were admitted to ICU.
- The logistic regression model achieved the highest F1-score of 0.493.
- Leucocyte counts and C-reactive protein were identified as significant predictors for ICU admission.
- The model demonstrated an AUC of 0.84 on the test set.
- 87.3% of ICU admissions were in women in the third trimester.
- Older age and more pregnancies were associated with higher ICU admission rates.
- Lower hemoglobin and lymphocyte counts were observed in ICU admitted patients.
- Anemia was frequently reported among those requiring ICU admission.
Takeaway
This study used computer programs to help doctors figure out which pregnant women with COVID-19 might need extra help in the hospital.
Methodology
A retrospective study analyzing data from COVID-19-infected pregnant women admitted to hospitals in Kazakhstan, using eight machine learning classifiers.
Potential Biases
Potential underrepresentation of more vulnerable groups and variability in treatment protocols.
Limitations
The study is retrospective, conducted in two centers, and may not represent all pregnant women due to unvaccinated status during the study period.
Participant Demographics
1168 pregnant women with COVID-19, primarily in the third trimester.
Statistical Information
P-Value
0.0011
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website