clinical study

Introducing Artificial Intelligence Applications at Our Community Hospital: A Contrarian Approach

The Problem

With artificial intelligence (AI), the challenge was which applications should be introduced first to get all the stakeholders (health system administrators, patients, clinicians and radiologists) interested in the AI culture and maintain their enthusiastic support

The Solution

We found three FDA-cleared AI solutions available in the marketplace that aligned with three clinical situations we wanted to improve. These included long acquisition time for PET/CT & MRI scans, and long wait times for the radiologist’s report on critical CT findings.

Outcomes and Limitations

AI for flagging the emergency department cases for ICH, PE, and LVO was introduced. Three months of validation of the AI software took place, with false-negative rate as the primary method of validation. No false-negative cases were found during the validation phase. A very small subdural ICH identified by the AI software may have been missed by the radiologist. The accuracy of the algorithms for PE and LVO is extremely high, and AI excels in identifying PE in the small vessels.

Conclusions

Relatively small subdural ICH identified by the AI software may have been missed by the radiologist.  The accuracy of the algorithms for PE and LVO is extremely high, and AI excels in identifying PE in the small vessels.

Read Study

Aidoc Ask the Experts Webinar

Popular Posts

Article
A&E and AI: How Everlight Radiology Deploys AI to Support NHS Trust Care Expediency
Article
Three Key Takeaways from the 2023 PERT Consortium
Article
Nine Essential Considerations When Choosing a Clinical AI Vendor 
Article
Fostering Innovation & AI at a Health System: Novant Health

Stay Connected on AI Trends and Insights

Follow us on social

From Our Blog

View all

© Copyright 2023 Aidoc