Skip to main content


AI in Healthcare: Care Delivery Use Cases

Hand holding drawing virtual lightbulb with brain

The Hype Around Generative AI Continues

Many healthcare leaders are wondering if (and how) generative AI, the shiny new tool, could drive value in their organization.

Our recent discussions with Chief Medical Officers, Chief Information Officers, Chief Medical Information Officers, and a VP over Nursing point to this: AI can provide a huge amount of value when it comes to care delivery (e.g., point of care).

You may notice that I mentioned AI there and not, more specifically, generative AI.

Frankly, several of the most interesting care delivery use cases do involve generative AI, but they are not the only examples. And any truly innovative approach shouldn’t self-limit based on what’s absolutely hot in the market place. (Looking at you, generative AI.)

Many provider CIO’s and CDO’s tell us that the revenue cycle side of the equation is already supported by a number of helpful AI solutions. But, they stress, that doesn’t solve for some of the most vexing problems when it comes to clinician burnout. So in this post, I’ll focus on use cases for care delivery and how AI can help.

Care Delivery Use Cases: AI (and Generative AI, Too)

These use cases are just that – a list of possible uses for AI/ML and predictive analytics that drive some sort of value. AI can do a lot, but every use case assumes that the AI model or tool can be used with a system of engagement like an EMR.

SEE ALSO: Evolving Healthcare: Generative AI Strategy for Payers and Providers

I’ll break each of the use cases down by category:

Accelerate Imaging Decisions

We’ve helped our clients identify multiple imaging use cases, including the ability to:

  • Flag radiology concerns: Use AI to review an image and perform a preliminary radiology modeling assessment. The focus is to help radiologists start with pre-identified concerns on a given x-ray, CAT scan, or MRI.
  • Identify patients who need specialty support: This is similar to radiology in that an assessment occurs, but it would be for pulmonary nodules, liver transplant, kidney transplant, etc.

Streamline Referral Processes

AI could help to better manage the referral process. Of course, it would have to be paired with engagement technology and capabilities (which we also drive for our healthcare clients).

  • Manage referrals: Receive a referral and apply an AI model to review referral data and determine the correct doctor or other component of the referral.  You can make this part of a referral management workflow which automates this step.
  • Improve experiences: You can also automate the actual experience with the patient. communicate via the preferred channel, send emails or text with the referral info, and allow the patient to interact with an AI chatbot to schedule an appointment.

Support Operations With Deep Learning (DL) Insights

Delays in hospital throughput have negative impacts on financial and hospital optimization results. Insights derived from AI could help close the gap.

  • Length of stay: Define potential length of stay in a variety of situations including pediatric, time study, overall patient population, etc.
  • Readmission risk: Define the risk of readmission and identify the most likely cause of readmission; this AI model should identify the risk and then push the insight to an EMR.
  • Left without being seen: Identify who will leave the ER before being seen.
  • PT/OT/OT: Determine which patient should be treated with therapies; this becomes an aid to help the clinician diagnose and prescribe.

Support Patient Care Decision Making

Well-crafted models could support teams as they make decisions to support patient care.

  • Testing decision-making: Define appropriate (and unnecessary) testing under certain conditions
  • Evidence-based clinical decision-making: again, this AI model will help define clinical decisions that a clinician can then use in their decision making process
  • Digital twins: digital twin data could be used to help define the overall best care for a patient

Auto Generate Medical Records

  • Using ambient clinical intelligence, capture conversations between a patient and doctor or nurse and then generate an encounter in the EMR with any needed prescription or other information. Ideally, the doctor would review and approve the record and any actions coming from the visit. (This approach should be used for any patient interaction, whether it be a doctor, physician’s assistant, nurse practitioner, or nurse.)

Streamline Clinicians’ In-Baskets

Most clinicians feel buried in their in-basket and need help to quickly identify what needs quick action and what can be delayed or even automated. AI and generative AI could support in a number of ways:

  • Better manage the in basket. Auto-categorize the message and make it easy for the important things to rise to the top. Auto-forward messages where appropriate. Auto-generate forms and filters to help in quick response to a given in basket message.

Support the OR

  • Optimize the schedule: Based on a variety of factors
  • Help with surgical correct counts: Help to automate or even review to ensure no foreign elements remain inside post surgery

Elevate Patient Charts

Every clinician reviews a chart before and after speaking with a patient. You can use AI models for variety of purposes:

  • Auto-flag risk profile or patient risk scoring
  • Auto compile quality data results (in many hospitals, this is still a fully manual process)
  • Auto-chart review to auto-populate frequent screens
  • Pull in SDOH predictors of delayed discharge

Support Quality Audits

  • Auto compile data audits. Take the manual process, feed data to a model and compile a quality audit. This is technically a generative AI use case but one not top-of-mind.

Ease Digital Interactions, Using the Digital Front Door

What you’ll quickly notice is that this category differs from the others.  It deals more with digital interactions either before or after the point of care. That said, AI can still impact care delivery for things like correct identification of issues before a patient arrive.

  • Use AI to schedule appointments. Remember that scheduling an appointment many times relies on a practitioner’s skills and knowledge before actual creating the appointment.
    • Is the physician seeing patients?
    • Is this a specialist who can only be scheduled under the right conditions?
    • Will the physician be in a location at the time the patient wants to schedule?
  • Self-triage with a suggested appointment time and scheduling options
  • Use AI in a symptom checker and pair it with a chat bot to push to an e-visit, if applicable

Better Outcomes for Patients and Clinicians

One thing you will notice is that care delivery use cases focus on two main needs: 1) clinician burnout – in essence, making a doctor or nurse’s life easier – and 2) better care for the patient – getting them to the right care more quickly. AI offers tremendous potential to create better outcomes for both patients and clinicians.

In my next post, I’ll focus on correct prioritization.

Expert Digital Health Services: Imagine, Create, Engineer, Run

Our healthcare and data experts can help you identify AI opportunities and build a pragmatic implementation plan that holistically considers data, technology, and people.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Michael Porter

Mike Porter leads the Strategic Advisors team for Perficient. He has more than 21 years of experience helping organizations with technology and digital transformation, specifically around solving business problems related to CRM and data.

More from this Author

Follow Us