AI-enabled Emergency Information System

The chief of the emergency department leads the AI team and alongside many senior attending physicians, proposes various AI projects based on clinical needs. The AI team develops these ideas into usable applications. In consideration of the busy emergency environment, AI products must be seamlessly integrated into the existing emergency HIS, which allows for easy operation and minimal disruption to daily routines. Automated reminders are given to clinicians. However, final decision-making still falls into the physician’s hands.

Building upon established medical literature, we’ve developed a series of AI prediction systems for emergency medical conditions. Currently, our AI automatically predicts the risk of severe outcomes for seven major diseases in the emergency department including: pneumonia, elderly influenza, dengue, hyperglycemic crises, chest pain, pancreatitis, and traumatic brain injury. These outcomes include acute myocardial infarction, septic shock, admission to hospital, admission to ICU, and death. Since the outcome of concern is different in these diseases, manipulation of data and application of machine learning and deep learning algorithms produce varying predictive models. Currently, these seven diseases predict the following outcomes:

Emergency AI disease prediction system

	Disease	AI Predictive Outcomes
1	Pneumonia	Respiratory failure, Septic shock, Death
2	Elderly Influenza	Hospitalization, Pneumonia, Septic shock, ICU admission, Death
3	Dengue	Septic shock, ICU admission, Death
4	Hyperglycemic Crisis	Septic shock, ICU admission, Death
5	Chest Pain	Myocardial infarction, Death
6	Pancreatitis	Septic shock, ICU admission, Death
7	Traumatic Brain Injury	Hospitalization, ICU admission, Death