Deep Learning in the Medical Domain: Predicting Cardiac Arrest Using Deep Learning.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Decision Support

Background:

• Electronic health records (EHRs) adoption aimed to reduce mortality but face challenges due to low accuracy and false alarms from current rapid response systems.
• Existing solutions like Modified Early Warning Score (MEWS) and random forest machine learning have proven ineffective in addressing these critical issues.

Purpose of the Study:

• To introduce DeepEWS (Deep learning based Early Warning Score), a novel deep learning algorithm designed to enhance the accuracy of early warning scores.
• To address the limitations of current healthcare alert systems and improve patient outcomes.

Main Methods:

• Development of a novel deep learning algorithm for an Early Warning Score system.
• Evaluation of DeepEWS against the standard of care and existing market solutions.

Main Results:

• DeepEWS demonstrates high accuracy, outperforming current standard of care and market solutions.
• The system maintains superior accuracy levels even when considering the number of alarms in a clinical setting.

Conclusions:

• DeepEWS represents a significant advancement in early warning score technology.
• The novel deep learning approach offers a more accurate and reliable solution for patient monitoring in healthcare.