Wellbeing Forecasting in Postpartum Anemia Patients.

Area of Science:

• Maternal Health
• Computational Medicine
• Public Health

Background:

• Postpartum anemia is a prevalent global maternal health issue impacting mood, energy, and cognition.
• Current clinical assessments for postpartum complications are often intuitive, with a significant gap before follow-up visits.
• Restoring iron stores is crucial, but early prediction of adverse outcomes remains challenging.

Purpose of the Study:

• To investigate the efficacy of machine learning algorithms in forecasting postpartum depression and fatigue in anemic mothers.
• To identify key predictive features for these maternal health outcomes.
• To assess the clinical utility of ML models in managing postpartum complications.

Main Methods:

• Utilized data from 261 patients to train machine learning models.
• Developed forecasting models for depression (Edinburgh Postnatal Depression Scale - EPDS) and fatigue (Multidimensional Fatigue Inventory - MFI).
• Employed elastic net regression and compared model performance against baseline predictions.

Main Results:

• Machine learning models significantly outperformed baseline models in predicting EPDS and fatigue scores.
• The elastic net regression model achieved a mean average error of 2.3 for EPDS prediction.
• Edinburgh Postnatal Depression Scale scores and tiredness indexes at birth were identified as the most significant predictive features.

Conclusions:

• Machine learning models show strong potential for clinical application in predicting postpartum depression and fatigue in anemic patients.
• This approach can enhance early detection and management strategies for postpartum mental health and fatigue.
• Integrating ML tools could improve patient well-being and streamline postpartum care pathways.