Machine Learning: Achieving a Streamlined Prior Auth Process

While the terms artificial intelligence (AI) and machine learning (ML) may conjure thoughts of robots running the world, the reality is that these techniques for automated learning positively impact many industries. In particular, given the vast amount of data and information within the healthcare space – and an industry-wide emphasis on streamlining technologies – machine learning presents an ideal opportunity to help simplify an increasingly complex ecosystem.

What is machine learning?

In traditional programming, programmers use a manual process to create a static code that helps drive to a specific outcome; the program does not change until that code is manually updated. In contrast, machine learning enables computers to learn from past experiences and data without being explicitly programmed. Fundamentally, ML uses math and statistics to build models about relationships. To better visualize the differences, think about cooking your favorite dinner. In traditional programming, you provide the ingredients (the input), combine them with the recipe (the traditional program) and get dinner on the table (the output). Alternatively, with ML, you provide a list of ingredients (input) and pictures of the final plate (output), and the computer reverse engineers the recipe (the program).

Traditional Programming

Traditional programming is a manual process – meaning a person (programmer) creates the program. But without anyone programming the logic, one has to manually formulate or code rules.

In machine learning, on the other hand, the algorithm automatically formulates the rules from the data.

Machine Learning Programming

Unlike traditional programming, machine learning is an automated process. It can increase the value of your embedded analytics in many areas, including data prep, natural language interfaces, automatic outlier detection, recommendations, and causality and significance detection. All of these features help speed user insights and reduce decision bias.

ML offers numerous advantages. For example, leveraging ML enables organizations to process tasks faster: computers can perform simple, repetitive tasks quickly, like finding a certain word among pages of text. ML also provides an advantage when it comes to quality – knowing that a computer is able to do certain tasks more consistently than humans – and lends itself to scalability by building reliable solutions. However, machine learning isn’t a solution for every problem. Effective ML systems shouldn’t replace people; instead, ML should bolster them by providing a support tool to improve workflows and unburden them of wasteful, repetitive tasks. 

Reshaping healthcare

With computers being taught to analyze vast amounts of data, it’s no surprise that machine learning is being leveraged within the healthcare industry, and with great results. AI and ML have contributed to many clinical use cases including decreasing the cost to develop life-saving medicines, forecasting kidney disease, and even providing assistance to emergency medical staff.* Until recently, however, UM programs and processes, specifically those related to prior authorization, have been slow to adopt these technologies.

At Cohere, we’re leveraging AI and ML to improve prior authorization. We incorporate ML into our platform using data from multiple sources, such as medical claims, clinical reviews, service requests, and clinical data derived from the EMR. Our ML principles, which help guide us along the way, include:

• Continuous releases that provide incremental improvements, agility to grow and change, and help mitigate risk
• Modular components that help us leverage flexibility for multi-use applications and phased implementation
• Process improvement through seamless integration with existing workflows, decision boosting through guidance, and automation where possible
• Transparency with explainable models and an open decision-making process 

Benefits of ML in utilization management for health plans

For health plans, the key to unlocking the value of machine learning for utilization management (UM) is recognizing that implementing ML in prior authorization workflows goes beyond simply digitizing legacy manual processes. This technology affords a much greater depth of value for health plans, with the capacity to drive considerable impact with regards to medical expense savings, reducing operating expenses, and improving the provider experience. While these goals may seem lofty, machine learning can guide the way. To reach these goals, health plans need technology capabilities that include

• Easily identifying and flagging clinically inappropriate cases for review
• Reducing the time it takes to review each case
• Improving turnaround times by increasing the number of cases that are automatically decisioned
• Accommodating a variety of submission channels
• Reducing the burden placed on providers by manual data entry 
• Identifying and targeting providers who are misrepresenting clinical information

Cohere harnesses ML to ensure health plans can effectively and efficiently work towards these goals. Cohere’s platform can identify and prioritize reviews based on clinical appropriateness and potential medical expense impact; launch ML-driven “nudges” that help steer providers to lower-cost alternatives before submission; help pre-fill clinical data where possible in order to streamline the service request process; and improve the quality of manual reviews so that opportunities for impact are not missed, among others. By implementing these techniques, we continue to drive positive provider experiences, get patients the care they need faster, and reduce administrative burden throughout the entire process.

Creating impact for the future

While machine learning has become a buzzword in the healthcare tech space, the reality is that it takes time, attention, and immense detail to develop, launch, and grow a successful machine learning strategy. Cohere continues to focus on ways to provide true impact across the healthcare spectrum, and we know that machine learning is one way to help get us there.

Download our white paper to learn more.

*Contributed: Top 10 Use Cases for AI in Healthcare, MobiHealthNews, 07/01/2022