AI in Healthcare: Predictive Analytics for Patient Outcomes
Abstract
This research investigates the impact of AI on healthcare workers reasoning skills due to AI based applications.
Methodology
- Data Source: MIMIC-III Clinical Database.
- Preprocessing: Imputation of missing physiological data, normalization, and sequence generation.
- Model Architecture: Bi-directional LSTM with attention mechanisms to weigh key clinical events.
Key Findings
- The proposed model outperformed traditional scoring systems (e.g., SOFA, APACHE II) by 15% in AUROC.
- Early identification allows for timely intervention, potentially reducing mortality rates.
Publications & Resources