AI adoption is on the rise, with a recent Gartner survey showing that 87% of those surveyed plan to deploy AI within the next two years. For health insurers, AI offers an opportunity to streamline often manual, labor-intensive processes as well as deliver better, more personalized member experiences.
As a result, health insurers have begun to explore and implement AI in earnest. With that in mind, here’s a brief guide for health insurers looking to implement AI.
What is AI?
At a basic level, AI is intelligence demonstrated by machines. AI essentially enables machines to reason, learn, plan and in some cases be creative.
There are several different flavors and subsets of AI. One of them, machine learning, is generally what people refer to when talking about AI. In machine learning, a computer system learns how to do a task instead of being programmed how to do it. Generally, machine learning requires large amounts of data that a machine can learn from. The majority of AI applications gaining traction today leverage machine learning to improve business processes or customer interactions.
Assess Your Resources
The first step when implementing AI at a health insurer is to analyze your internal resources. Do you have the requisite human and technological resources on staff to implement your AI strategy? Is the required data available and is it structured? Do you have the technology infrastructure you need to implement your AI strategy?
If not, you have options. You can take a DIY approach and try to build those resources. That will take time, but like most build vs. buy decisions, deliver more flexibility and embed AI deeper into your organization. If you decide to buy, there are many AI consultants and other resources available to help you build AI into your business.
There are also multiple ways to complete an AI project, moving along a continuum that includes building a solution from scratch, buying off-the-shelf products and functions, collaborating with a partner, or outsourcing the development completely. There are pros and cons to each of those, but a lot of that decision-making process revolves around what existing AI resources exist, how much business impact your use case has, and how quickly you’d like to go live.
Ideate Use Cases
Speaking of use cases, a good next step is to identify AI use cases. Some things to look for when creating a list of AI use cases:
- Data-based — Generally speaking, use cases most prone to AI automation are already data-based. That means any decisions made are completed using data.
- Frequently Occur — Repetitive decisions based on data that consume a lot of time are ripe for AI.
Review our 7 AI use cases in Health Insurance post for more information, including examples of how other health insurers have leveraged AI.
Prioritize Based on Strategic Objectives
The best way to get executive buy-in for an AI project is to show how it aligns with your organization’s strategy and goals. For example:
- Improve NPS Scores — If your organization is trying to transition to a member-centric organization and highly values NPS scores, any member interaction that is commonly rated poorly by members is a prime candidate for improvement through AI. Some organizations leverage AI to match members with the member services rep best available to deliver a great experience. Or, you can leverage AI to perform member outreach based on claims data. For example, sending members who recently experienced a surgical procedure tips about how to reduce scarring.
- Reduce Administrative Costs — If your organization aims to reduce administrative costs — and which don’t? — claims processing automation may be a great AI use case. Or, you may investigate ways to reduce manual intervention in the prior authorization process.
Whatever the use case, tying it to a business objective can help prove the value of AI in your organization. Even more important — make sure you describe your use cases as clearly as possible. Consider outlining the problem solved, the value proposition, estimated ROI, and key success metrics at a minimum.
Start Small and Iterate
Often, tuning AI models and algorithms can take time before you experience tangible results. For that reason, proving the value of an AI project requires you to set milestones along the way. Start small and build basic projects first. As you gain experience — and success — you can undertake more complex projects or build more elaborate algorithms.
Similar to the Agile software development methodology, it’s important to break large projects into smaller chunks and iterate over time to learn and grow your AI expertise.
Collaboration is the key to developing successful AI projects. A successful AI project usually requires resources from across the organization, including:
- Business Operations Knowledge — If you’re building claims processing AI solutions, you’ll need expertise from the team responsible for processing claims. Their subject matter expertise is vital when developing a working AI solution.
- IT Knowledge — You’re also going to likely need resources that understand your business data, like the data engineers on your IT team.
- AI Knowledge — Finally, you need team members with AI skills to create AI applications.
With such a diverse team, you’re going to need to encourage communication within the group, but also out to the business. For many organizations, that means engaging a member of the leadership team to champion AI projects.
Finally, celebrate successes. You may not experience $260 million in savings like Highmark did when implementing fraud detection techniques that included AI, but highlight your successes internally. As you build more successful AI projects that save the organization time and money, more employees and leaders will think about ways to implement AI within their job or team.
One significant aspect of celebrating successes is setting goals. For example, if you are building an AI chatbot to improve member access to data, set a goal for that AI project. You may measure that through a reduction in member support calls related to specific information. Or, you may survey members and set a goal to increase the average response to a certain question. Whatever the case, having metrics that help show the impact of an AI project can help other team members understand the impact of the project.
Certifi’s health insurance premium billing and payment solutions help healthcare payers improve member satisfaction while reducing administrative costs.