With all the scary changes in the world, I’m committed to sharing some of the great news about AI. I recently wrote about advances in educational technology: Today, let’s look at an exciting use of AI in healthcare.
The Challenge
The University of Texas Medical Branch (UTMB) is a perfect example of what’s possible when medical expertise meets innovative technology.
The toxicology department screens biological samples for the presence of drugs, alcohol, medications, and other potentially harmful substances.
The Challenge:
UTMB’s pathology team faced a bottleneck:
- Processing only 50 toxicology cases per week
- With each case requiring 15 minutes of manual review across 40+ metrics.
This slowed their ability to meet growing patient demand.
The Solution
UTMB developed AutoTox., which is an AI-powered diagnostic tool built on the Retool platform. AutoTox integrates with existing systems and uses AI to generate simplified reports for pathologist review, streamlining the workflow.
The Results:
- 66% reduction in testing time (from 15 to 5 minutes per case)
- 10x increase in capacity (from 50 to 500 patients screened per week)
- Increased accuracy by helping pathologists spot potential errors they were more likely to miss during a manual review process
Human in the Loop
Amazingly, UTMB’s Chief AI Officer Dr. Peter McCaffrey, MD, MS, FCAP built the entire application in just 5 weeks. Dr. McCaffrey says:
Some critical detail about this app in particular is that it is human-in-the-loop. Core to this value proposition is that the concept of “Review time” isn’t strictly time spent thinking about the implications of the tox data for the patient. As time studies routinely show, much of this physician time for case review and patient review overall is tied up with non-clinical non-top-of-license activities like searching for the active medications, LCMSMS results, and relevant demographic or historical factors. Overall review time goes down not because less time is spent on thinking about the meaning of the tox results but less time is spend on those OTHER things.
Dr. McCaffrey also talked about RVUs, or Relative Value Units, which is a measurement used in the healthcare system to measure the value of healthcare services. He says:
It’s also worth considering the implications of that kind of automation on the RVU pricing for this–and many other–activities. Since RVUs broadly capture a concept of “complexity” and “time spent” they do also include these search, retrieval, and aggregation activities. As these become more automated, a Pathologist may actually spend more time thinking about the patient case but still less time overall turning over the case. The result may be both an improvement in quality and a potential reduction in RVUs.
Augmenting Human Expertise
In this particular case, AI is augmenting human expertise, not replacing it. Now pathologists have more time to focus on complex cases, and patients don’t need to wait nearly as long for answers.
What excites you most about AI’s potential to help people solve real-world problems?