FREE CASE STUDY
Know before you buy.
Start now by downloading our free case study to learn how our comprehensive approach to validation helped a leading radiology practice identify the best AI tools for their patient population.
Know how your AI Performs
94%* of FDA-cleared algorithms perform differently on real world patients.
Ferrum ensures your AI initiatives meet your clinicians’ standards by continuously validating and monitoring algorithm performance on your actual patient population.
Validate
Run algorithms retrospectively on your own patient data. Quantify the clinical and business impact of new tools and reveal where they under- and over-call.
Monitor
Ensure continued performance of your algorithm library as your practice patterns, patient populations, and technology stack changes.
Update
Track how your algorithms improve over time and across updates. Identify new blind spots and failure points before they affect patient care.
* Based on data from External Validation of Deep Learning Algorithms for Radiologic Diagnosis: A Systematic Review. Learn More
Application Validation Continuum
Validation and Monitoring
The application validation continuum extends from the extreme of having no validation at all all the way up to having fully automated validation processes.
- Random selection of studies pulled from PACS
- Manually labeled by resident or junior staff
- Requires huge volumes of studies labeled for low-incidence conditions
- Requires huge volumes of studies labeled for low-incidence conditions
- Minimum radiologist time required; language Classifier results establish ground truth
- Able to build large data sets of low-incidence conditions (ie. pancreatic cancer)
How Continuous Validation Works
Ferrum continuously validates AI tools by running visual classifier algorithms in parallel with language classifiers (based on specialist reports) and then comparing this results. This process results in continuous improvement in the accuracy of the results.
Working with Ferrum Health has been brilliant. We’ve been able to deploy multiple AI algorithms in different clinical settings, and push data to various clinical workflows. The impact for our radiologists, technologists and patients has been positive and significant. The Ferrum Health team has done most of the heavy lifting and have been hugely responsive to our feedback for improvements.
Dr. Lloyd McCann
CEO
