Healthcare is moving faster to adopt AI than any other industry; the most effective implementations remove repetitive and rote work
AI can do amazing things, yet the technology is prone to making mistakes. Spotting those errors requires expertise.
Mark Bates, who leads the product team here, summed this up nicely for Healthcare IT Today:
“The best use case is augmenting existing skills, knowledge, and experience, especially in a nuanced industry like healthcare. While generative AI can process and organize vast amounts of information – beyond a human capacity – it still takes a layer of institutional knowledge to identify the kernel of insight worth pursuing.”
That’s the approach we’ve taken in adding AI to our rural healthcare recruiting platform. It’s also the approach that seems to be most pervasive in healthcare.
A recent story by The Wall Street Journal (WSJ) – Hospitals are a proving ground for what AI can do, and what it can’t – broadly reflects this point. “Doctors still make medical decisions, though AI can aid the process,” wrote authors Te-Ping Chen and Chao Deng.
They go on to highlight several use cases, including:
- Recognizing “subtle signs of breast cancer” between screenings;
- Helping doctors to spot more “growths during colonoscopies”;
- Reviewing a million medical images and flagging a few for a closer look;
- Analyzing medical charts to prioritize patients by risk across “21 hospitals”;
- Providing “tailored nutrition advice” for “breast cancer patients”;
- Checking the “latest standard of care” for irregular “conditions”;
- Mining patient records to automatically draft claim appeals; and
- Transcribing patient notes.
Many of these use cases center on repetitive and rote tasks. Automating them frees up a provider’s time to focus on those requiring greater cognition.
In other cases, it calls attention to signs a provider might have missed. For example, the time and effort a human would need to review a million images is prohibitive. Yet the AI narrowed it down to a manageable list of 70. When humans looked at that refined set, they found “five instances where physicians deemed that more follow-up was needed,” the WSJ reported.
Other possible use cases for AI in healthcare
Healthcare is adopting AI at a rapid pace. An October 2025 survey by Menlo Ventures – cited in the aforementioned WSJ article – found that, “In just two years, healthcare went from 3% adoption to…27%, outpacing outpatient facilities (18%) and payers (14%).”
With so many organizations experimenting with it, we’re bound to see more and more creative applications of the technology in healthcare settings. Below is a summarization of AI use cases we’ve seen reported recently.
1. Automate prior authorization
A survey by Deloitte found 93% of “health plan executives” expect AI to “ease” prior authorizations, Becker’s reported. AI could conceivably mine patient records to “validate clinical criteria.” If this worked well, it would bring substantial relief to burnt-out clinicians who frequently cite prior authorizations as one of the biggest hassles of working in medicine.
2. Assistance in completing assessments
Vivid Health is developing a product to help at-home health service providers fill out “assessment forms” more easily, according to Fierce Healthcare. Existing solutions reportedly don’t meet the needs of at-home providers.
Patient intake, particularly the Medicare Outcome and Assessment Information Set (OASIS) form, is an onerous form to complete. “Using Vivid can double daily patient intake, reduce nurse paperwork by 75% and reduce intake time to 30 minutes or less, the company touts,” the article says.
3. Improving radiology workflow
Radiologists must bounce around various tech tools in reviewing images, archiving them and storing data. A company called RadNet is developing an “AI-powered remote scanning platform” that provides a multi-modal review that enables collaboration, says a report in Health Data Management.
The cloud-based product is “vendor agnostic.” This means the task list, “reporting, viewer, advanced visualization and analytics … can be synced together to enable continual improvement and build trust in AI.”
4. GPT dedicated to healthcare
OpenAI launched ChatGPT for Healthcare. A report by HIMSS’ Healthcare Finance says the product provides answers “grounded in medical sources from peer-reviewed research studies, public health guidance and clinical guidelines, with citations given for quick source-checking.”
It’s also been “evaluated through physician-led testing across benchmarks and real workflows” to assuage concerns about response integrity. Several health systems are using the product. Among them are AdventHealth, Boston Children’s Hospital, and Cedars-Sinai Medical Center.
5. Follow-up phone calls to patients
Universal Health Services and Hippocratic AI developed an AI agent that helps providers follow up with patients after discharge, according to Healthcare Dive. The AI agent will call “patients to review medication instructions, probe for signs of new or worsening symptoms and answer patient questions.” It also provides a way for patients to get a call back from a nurse.
6. Checking patient eligibility
Healthcare Dive also noted Salesforce rolled out agentic tools last year for providers and payers to reduce “time-consuming tasks.” The product has a range of use cases and one that stood out to us was “patient access and services capabilities.” It helps answer questions and check “patient eligibility with insurers.”
7. Synthetic health data for training AI
Training data is among the biggest challenges for AI – there simply isn’t enough of it. In healthcare, the problem is compounded by strict privacy rules. A journal article in Science Direct highlights the GANerAI architecture as a possible solution.
It creates tabular synthetic data that’s often used in clinical trials. The idea isn’t new, but the approach is, and it’s markedly better, according to the authors’ conclusion:
“While other networks exhibit both long training times and less than optimal data generations performance, our newly introduced GANerAid architecture seems to produce more satisfying synthetic patients that closely resemble original data, as shown in the experimental results.”
* * *
See the software for yourself! Healthcare Workforce Management is designed specifically for rural healthcare recruiting: Contact us for a no-obligation demo.
If you enjoyed this post, you might also like:
Lost billings: The financial impact of provider credentialing process in healthcare
