MDLIVE for Cigna cuts wait times by 50 percent by predicting changing healthcare demands with Azure Machine Learning

Telehealth provider MDLIVE for Cigna serves 62 million people in 50 US states and 10 territories. It wanted to improve its forecasting to reduce patient wait times and balance workloads. The company worked to maintain high-quality services without overburdening staff during the COVID-19 pandemic. It started collaborating with AIDAN Health (AIDAN), a Microsoft partner, to develop machine learning models that could improve its operations.

AIDAN used a many models forecasting architecture on Azure Machine Learning to meet MDLIVE for Cigna’s needs. The company also used Azure DevOps to automate and iterate rapidly, enhancing teamwork. “The collaboration with AIDAN has been very good,” says George Regnery, Chief of Staff for MDLIVE. “We’ve gone back and forth with them to continually improve the model, and it keeps getting better.”

With its new, fine-tuned models, MDLIVE for Cigna greatly improved its load balancing. It has reduced patient wait times by more than 50 percent even as usage has grown each year. Now the company is better equipped to attract and retain qualified medical professionals.

Forecasting through COVID-19 pandemic difficulties 

MDLIVE is a provider-led telehealth service founded in 2009. After the COVID-19 pandemic upended the operations of many healthcare organizations, Cigna acquired MDLIVE as part of its pivot to telemedicine in 2021. “All the familiar patterns broke down in March of 2020, when visits skyrocketed,” says Regnery. “We were faced with the conundrum of forecasting in the wake of the COVID-19 pandemic.” 

Labor market dynamics added another layer of complexity. MDLIVE for Cigna’s providers are independent contractors, and the company wants to make sure they are happy and not overwhelmed with their workloads by hiring more staff during busy seasons. It takes 90 to 120 days for MDLIVE for Cigna to verify the credentials of its providers, so it must have a way to predict future demand if it is to meet patient needs. Moreover, licensing requirements demand that physicians maintain active licenses in each state where they provide care, even if they are only seeing a patient online, which adds to hiring difficulties. “We need to forecast well in advance to ramp up for peak season,” says Keith Bergquist, Chief Operations Officer for MDLIVE for Cigna.

Halving patient wait times with Azure Machine Learning

All healthcare providers strive to eliminate wait times. However, very few have adopted advanced analytics to forecast patient demand. MDLIVE for Cigna tackled this industry-wide challenge in tandem with AIDAN. Previously, the telehealth provider was creating forecasts only once a year. The new solution has changed that. “By having the forecasts on a weekly and monthly basis at the state level, we can evaluate our provider numbers and solicit more if we need to,” says Bergquist. However, that fine-grained approach required a cutting-edge solution.

MDLIVE for Cigna started working on this project with AIDAN in late 2021. AIDAN chose to develop forecasts on Microsoft Azure to accommodate the national scale of MDLIVE for Cigna operations. AIDAN also built models that took into account different variables for every state and territory. Some, like time of year, were related directly to seasonal illness. However, others, like unemployment rates, impact healthcare usage indirectly by affecting health insurance coverage. In order to accommodate so many variables, AIDAN used a many models architecture. “There was a big iterative process that made our model more accurate,” says Regnery. “We kept adding granularity into the model.”

The companies were developing the model even as the healthcare industry continued to experience historic changes. First, different strains of COVID-19 swept through the population, and then a series of respiratory infections drove up healthcare usage. “In 2022, we saw about 25 percent higher patient volume than in 2021,” says Bergquist.

The project with AIDAN was completed in spring 2022. Since then, the forecasting model has helped MDLIVE for Cigna in many ways. One such way was with provider availability. Previously, the company had to offer monetary incentives to boost doctor availability during peaks. But with the new models, that hasn’t been necessary. “We’re saving about $1 million each busy season with our Azure Machine Learning models,” says Bergquist. The improvements have also enhanced the patient experience. Since implementation, wait times have fallen by more than 50 percent. After submitting their request to see a doctor, patients are only waiting about 20 minutes to receive a phone call. And MDLIVE for Cigna is seeing higher patient volumes than ever before. “In November 2022, we served 40,000 more patients than in November 2021,” says Bergquist. 
 

Continuing to improve forecasting accuracy with AIDAN

Looking back on their success, MDLIVE for Cigna emphasizes how important it is for healthcare organizations to have advanced forecasting. “No matter what your business is, the ability to accurately predict demand is critical to running an effective operation,” says Bergquist. “We’re seeing a lot of business value from the model that AIDAN has built.”

Although MDLIVE for Cigna is happy with the results of its project, the work is not over yet. Patient demand continues to fluctuate in response to both seasonal and economic trends. The company sees its engagement with AIDAN as an ongoing relationship that will continue to require input from both companies as the market changes. “We like the way our model on Azure works,” says Regnery. “It’s given us more insight into the coming months.”