Good or bad, we all have to deal with insurance.
However, unless you work in healthcare, you've probably never encountered the appeals process. This process involves negotiations between insurers and healthcare providers to determine final prices.
Project Overview
When DocVocate hired my team at Kunai, all they had was an idea and an AI model. They needed us to create an entire product. The idea was to use AI to speed up the time it takes to write and submit an appeal and to improve the approval rate.
Customer Research
To get started, I needed to learn as much as I could about insurance claims and appeals, and just how problematic the process was.
I spoke with 25 medical providers, from small acupuncture clinics to large hospitals, observing how they managed their piles of rejected claims. They processed each appeal one by one, starting with the oldest. For each appeal, they retrieved medical files, pulled notes, scanned any tests or x-rays, and then typed a letter explaining why the insurance should cover the suggested amount. Each appeal took between 20-40 minutes.
From this research, I identified three key areas that our tool needed to address:
- Automating repetitive tasks
- Improving prioritization
- Creating a digital repository for medical files
Building the Product
![[Fig. 1] Flowchart of the insurance claim process by Appealio](https://cdn.prod.website-files.com/5e8613164ab55c0f2b687372/66196f3da627bdc955899c81_Appealio%20Flow.webp)
The Appeal Letter
Drafting this letter was the most time-consuming part of the process. We addressed this by programming the AI to gather data about the patient, the appeal, the insurance, and the procedure. It then used templates to draft the letter.
We designed a split-screen layout for ease of use: on the right, a text editor with the letter and data ready to edit, and on the left, a preview of the final letter.
![[Fig. 2] Split-screen appeal letter with text editor and preview.](https://cdn.prod.website-files.com/5e8613164ab55c0f2b687372/5e8de96b778337bfb09bd0b4_Letter2.png)
Prioritization
Working on appeals from oldest to newest led to missed deadlines and lost revenue. So, we created a better way to prioritize appeals. We developed an algorithm based on deadline, monetary value, appeal round, and chance of success to determine the most high-impact appeals. This helped medical providers generate more money from these appeals with a much higher success rate.
![[Fig. 3] Appeals table showing the list of appeals and the ability to sort and filter them.](https://cdn.prod.website-files.com/5e8613164ab55c0f2b687372/5e875942801f486621055928_Dashboard%201.png)
HIPAA and Document Storage
Storing medical data is tricky. To make the appeal process easy, we needed a system that allowed healthcare providers to upload records, test results, and images within our tool. It had to be secure and private. We followed all e-PHI (electronic Private Health Information) and HIPAA rules in the application and added encryption and other protections for the data.
![[Fig. 4] An appeal flyout showing data stored in the appeal (images, letters, past appeal information)](https://cdn.prod.website-files.com/5e8613164ab55c0f2b687372/5e8df3491604b7be4c22ab5e_Appeal%20Details%20History%20Copy.jpg)
Outcomes
We introduced the beta version to a select group of hospitals and small healthcare providers. The feedback from customers surpassed our expectations. The tool not only saved time and reduced stress but also increased financial returns for providers. They earned more by processing appeals faster and more efficiently, with fewer rounds of appeals and a much higher acceptance rate.
