Book a Demo

Identify patients who will no-show their doctors' appointments

When patients no-show their doctor’s appointments, it causes problems. The patients don’t get the care they need. Other patients could have been seen. Doctors and medical staff are unhappy about that and the hospital loses money.

Medumo helps hospitals solve this problem by guiding patients towards the care they need and making sure they show up for their appointments. As an analyst at Medumo, I’ve used Intersect Labs to predict the likelihood of a patient no-showing or canceling their appointment at the last minute, four days in advance of the appointment — giving the hospital time to react.

Because patient information is confidential under HIPAA, I have limited access to demographic information that might predict those risks, so I had to get creative. Fortunately, with behavioral analytics, we’re able to track how patients use the app, and that can help predict whether a patient will show up for their doctors’ appointment.

Pulling the historical training data was simple: we simply got a readout of the events each patient performed in the app, and counted them. The count of each event being performed became a column. Then, we joined that information with some basic information about the patient, and added a column indicating whether the patient showed up or not.

Each row represents an anonymized patient’s behavior, general location and complaint, and whether they showed up or not.

At first, I tried modeling this myself in R. I had to spend a lot of time cleaning and normalizing the data, properly encoding categorical variables, and then tuning the models. It was educational, but definitely time consuming. When I got the model working, I had to make the predictions locally — on my laptop. Deploying it into our app was its own engineering challenge.

When we found out about Intersect Labs, things got a lot easier. I didn’t need to spend any time cleaning the data or tuning the models: I only had to upload the raw historical data and tell it which column to predict.

Instead of coding for hours, I can make a lot of models quickly, test them, and they just work. They’re automatically deployed on a secure cloud, so we can easily make predictions via API and integrate them directly into our app.

Now, we can identify patients who are at risk of no-showing their appointments, and intervene to get them the care they need.

Interested in using Intersect Labs for your business? Schedule a free consultation call!

Case Study written by
Rohit Singh, Director of Analytics
Rohit Singh, Director of Analytics