A Guide to the Use of AI in Healthcare: Deploying a Realistic Approach

By MacKenzie Gonnelly

According to Stoltenberg's 11th Annual HIT Outlook Report, artificial intelligence (AI) and machine learning ranked as the top health IT industry focus across the next year. With AI's potential to augment patient care — through predictive analytics and modeling, improved diagnosis and treatment plans, and optimized care efficiency — the healthcare industry is eager to reap the benefits of this advancing technology. Yet still, healthcare is still in a discovery phase, identifying which methods of AI selection and deployment yield the greatest results. Despite AI's transformative promise, healthcare leaders must actively move beyond industry hype and determine the most practical applications for this technology. Consider the following four guidelines to establish a realistic framework for AI utilization within your hospital or health system.

1. Assess current organizational challenges
   While the healthcare industry has historically been slow to adopt new technologies, AI's potential has created a sense of urgency to implement this technology. However, when developing an AI program, HIT executives must first evaluate top organizational obstacles to pinpoint where AI might have a tangible impact. While AI tool selection will ultimately be unique to each organization, it is helpful to understand common healthcare challenges that AI is beginning to solve, such as:
   
   
   • Administrative burden: Time spent on administrative tasks detracting from direct patient care. After-hours admin time often leads to low job satisfaction, clinical burnout, or employee turnover.
   • Patient experiences: Evolving consumer-driven expectations push for more seamless, convenient, and efficient healthcare experiences.
   • EHR/ workflow inefficiencies: Inefficient EHR system workflows often lead to issues like end user click fatigue, burnout, and patient safety concerns.
   
   Assessing and prioritizing organizational pain points — and considering whether AI might be an appropriate solution — will initially guide hospitals and health systems in their decision making.
   
   
2. Start small and expand on healthcare AI solutions later
   According to a recent study, the number of healthcare AI product options are expected to expand fivefold by 2035. When considering numerous options, a best practice recommendation for HIT executives is to start with a modest approach, and gradually scale in size or scope. By first referencing current use cases across the industry, organizations can select tools that have been proven to work in the healthcare space. Notable AI solutions to the aforementioned industry challenges include:
   
   
   • Reducing administrative burden by introducing: Ambient clinical documentation, natural language processing (NLP) for EHR data input, or after visit summarizations powered by AI.
   • Boosting patient experiences by introducing: Personalized patient support tools like interactive chatbots or SMS/texts, automated medical coding or billing or maximized data insights.
   • Optimizing EHR systems and workflow by introducing: AI integration directly into the system, automated data entry or patient records management, augmented clinical data parsing for greater diagnostics support, or AI powered treatment personalization.
   
   Ultimately, HIT leaders should take peer reviews or recommendations into consideration when making purchasing decisions. Selecting established tools will allow AI's value to be more easily proven to various stakeholders, likely generating buy-in for custom solutions or additional AI utilization in the future.
   
   
3. Align AI tools with strategic goals and desired patient outcomes
   While the healthcare industry has historically been slow to adopt new technologies, AI's potential has created a sense of When considering various AI solutions, health organizations must also evaluate their current business or clinical expansion plans, determining how each tool might effectively support these objectives. Key stakeholders will likely review how AI will be integrated into an organization's technology roadmap or digital strategy.
   
   Healthcare organizations can achieve this alignment through several early-stage best practices. Firstly, it is important to remember that AI will never operate in a silo. In fact, AI should function alongside other tools or seamlessly integrate into existing systems for workflow and process efficiency. This will require leaders from across an organization to be on the same page throughout the AI implementation process, ensuring technology will not further burden clinical or operational teams. It will also require IT executives to balance AI initiatives alongside alternative projects and operational goals. Consider employing supplemental resources to alleviate a support burden for areas such as EHR support or maintenance, allowing internal IT teams to champion practical AI integration.
   
   Secondly, be prepared for the evolution of this budding technology by utilizing AI tools within existing vendor partnerships, including your EHR vendor. Because partnerships are already in place, opportunities for AI collaboration will naturally expand, without significant risk. Lastly, it's crucial to communicate to end users that skilled individuals are still essential for overseeing, operating, and managing AI. Consider adopting terms like “augmentation” within your organization, emphasizing that the goal is to support individuals, not substitute them. Also, prioritize the reduction of cognitive burdens for both clinical and non-clinical staff when implementing AI. Not only will end users be more inclined to adopt this technology, but the solution will garner maximum utilization for greater ROI.
   
   
4. Evaluate and combat potential risk associated with AI
   Although there are many opportunities for AI in healthcare, healthcare organizations must learn how and where the benefits of these solutions outweigh the potential risks. One major risk is the cost associated with this relatively expensive technology. According to a brief from IDC, more than 88% of IT leaders plan to increase their investments in third-party technology this year. However, all AI solutions must prove ROI — fairly immediately — to deserve budget allocation. Cost may be justified if AI tools provide robust evidence of increasing total patient volume, negating operational burnout or reducing costs associated with employee turnover.
   
   With the direct-to-consumer availability of many AI technologies, patients will expect this technology to be available in healthcare as well. However, according to the Journal of American Medical Information Association (JAMIA), concern has also been expressed about the ethical and regulatory aspects of the application of AI in healthcare — including data privacy and safety issues. To combat these risks, healthcare organizations must establish a strong governance structure to vet proposals and potential AI solutions. JAMIA proposes a four-step governance model — which includes fairness, trustworthiness, transparency, and accountability — that healthcare organizations must consider when developing AI programs.
   
   

While heath organizations may be eager to solve a host of operational inefficiencies through promising AI technologies, it is important to first establish a practical framework for AI utilization. By assessing organizational challenges, taking a modest approach, aligning tools with strategic goals, and combating risk along the way, healthcare organizations can set themselves up for success as AI continues to evolve.

For additional insights on AI's expansion into healthcare, visit the Stoltenberg blog.