AI in Healthcare IT: Innovation & Strategy Blueprint

Demystifying AI in Healthcare: Moving Beyond the Hype

New tech companies are developing specialized tools for every industry, and each one touts surges in productivity, capabilities, and efficient workflows. But there’s a lot of hype behind these claims, and buying in on the wrong tools can balloon your IT spending without significant ROI. Even worse, AI in healthcare can pose real risks that are more dangerous than unfilled promises. Data security breaches, “black box” methodologies that can’t tell you how AI tools arrived at different decisions, and unpredictable biases. Whether you’re exploring machine learning or generative AI, the only way forward is to balance innovation with a heavy dose of ethics and compliance efforts.

The Spectrum of AI: From Machine Learning to Generative AI

“AI” is a very broad category that encapsulates a wide range of potential tools. From image processing AI that can spot the first warning signs of cancer in millions of different patient images to smart AI assistants that help doctors and nurses intake patient data, “on the ground” tools can perform many different functions. Billing, procurement, and administrative tools can also help organizations trim waste, stay on top of inventory, and manage large swathes of personnel, patient, and medical data.

While it’s important to be cautious, it’s also important not shy away from AI adoption or to ignore the transformative capabilities it already houses. Both the cautious and the innovative can both agree on a fundamental starting point: building out the technical and governance infrastructure to make AI exploration safer and more implementation-ready.

Use Case #1: Enhancing Clinical Intelligence and Decision Support

The first use case is those “on the ground” applications. AI tools can manage dozens of small, repetitive tasks that bog down operations at clinics, hospitals, and doctors’ offices. For example, different tech platforms and programs could:

• Predictively analyze patients’ or individuals’ conditions over time to predict outcomes and interventions
• Analyze millions of data points to make initial diagnoses or verify staff’s decisions
• Develop personalized treatment plans for physician approval
• Take notes, summarize information, and send data to next-stage destinations
• Verify staff decisions against the latest available information regarding guidelines, drug databases, and regulatory information

Use Case #2: Streamlining Operations and Automating Administrative Tasks

The second, corporate-level use case is just as important (though just as perilous without proper implementation). Consider applications like:

• Generating, storing, and synthesizing notes
• Managing EHRs by verifying and cleaning data
• Verifying billing and coding accuracy
• Identifying trends in patient care and expenses
• Supporting efficient procurement and inventory management

The Governance Imperative: AI Security, Ethics, and Compliance

Governance protects your organization, your staff, and your patients from the potential hazards AI tools can bring—not just in erroneous “hallucinations” and widespread errors, but also in ensuring the data each tool accesses complies with HIPAA and other constraints. The four core tenets for good governance are accountability—ensuring there’s a clear line of responsibility, transparency—ensuring stakeholders can see how AI makes its decisions and what its activities are, fairness—that the AI doesn’t use algorithmic biases or bad training data that worsens outcomes, and safety through continuous monitoring and stringent screening of AI tools before they’re adopted. Humans must remain in the loop and in charge at all times.

The Platform Question: How AI Is Reshaping Core Systems (EHRs)

The first place where your organization might adopt AI is in your EHR. AI tools integrated into your preferred EHR—whether they’re native or secure third-party tools—can automate routine tasks, support documentation like a scribe, and summarize patient data based on a physician’s focus. AI can even generate smart checklists and prompt physicians if it notices a crucial step is incomplete. From small, optional nudges to active partnership, AI can transform EHRs from databases into assistants.

Building Your Organization’s AI Roadmap

Build your organization’s roadmap to incorporating AI with these broad steps:

1. Establish the goals of AI implementation.
2. Conduct a readiness assessment that evaluates data and technical infrastructure, as well as stakeholder buy-in.
3. Identify low-risk use cases for limited exploration and pilot programs.
4. Select secure vendors and implement tools more broadly.
5. Continue to scale and implement a concrete governance framework.

Take the Next Step in Your AI Journey

Virtually every healthcare organization benefits from adopting AI, especially in their EHRs and administrative functions, where repetitive tasks and large amounts of data open the door to human error. But it’s important to start with a partner that can help you identify your goals, build out your technical infrastructure, and ensure continuity of care as you experiment. Reach out to HPG to see how we’ve helped organizations like yours get started.

 

Image Credit: khunkornStudio/ Shutterstock