Hospital crowdedness evaluation and in-hospital resource allocation based on image recognition technology

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Hospital Management

Background:

• Hospitals face challenges in resource allocation, congestion, and patient experience.
• Current AI applications in medicine primarily focus on diagnosis, with limited use in management.
• Effective medical resource management requires addressing hospital congestion and optimizing patient flow.

Purpose of the Study:

• To investigate hospital congestion and medical resource allocation using artificial intelligence.
• To analyze spatial and temporal factors contributing to emergency department crowding.
• To identify areas of inefficiency and potential resource waste within a hospital setting.

Main Methods:

• Utilized image recognition via convolutional neural networks to process videos for real-time waitlist numbers.
• Defined a congestion rate based on psychological and architectural principles to measure crowdedness.
• Calculated diagnosis and post-diagnosis times from visit records, analyzing factors related to congestion.

Main Results:

• The pediatric waiting area exhibited the highest congestion rate (2.75) with 10,436 person-time.
• Pharmacy utilization was low (average 3.8 people at one time) with a low congestion rate (0.16), indicating space waste.
• Diagnosis and post-diagnosis times were linked to patient age, diagnosis count, and disease type, with respiratory issues being most common (54.3%).

Conclusions:

• The emergency department studied experiences significant congestion and medical resource waste.
• Artificial intelligence offers an effective tool for investigating and quantifying hospital congestion.
• Combining AI and traditional statistical methods can improve research on healthcare resource allocation.