Blog: Sleep & Respiratory Healthcare

Predictive Analytics Using Artificial Intelligence (A.I.)

[fa icon="calendar"] Jan 10, 2019 2:29:53 PM


The healthcare industry has built a goldmine of patient data but it's not being fully leveraged. In this video, watch Dr. Raj Misra - our Chief Data Scientist - discuss how Somnoware uses A.I. for predictive analytics. 162.png(Click the red button above to watch this video) 

In the healthcare industry, we are sitting on a goldmine of data, ever since the early 2000s when electronic health records (EHR) became mandatory and we started collecting a lot of data on every patient. The unfortunate reality is that nobody is really doing much with that data.  

At Somnoware we are trying to change that. Over the past decade and a half, data science and machine learning as disciplines have evolved quite a bit. Today, we are much smarter about analyzing data—looking and finding patterns in the data and about leveraging the power of data to help our stakeholders make better decisions. So, in this case for example, physicians can now look at the output at some of our data analysis and know how their patients are doing.

One of the projects we have been working on at Somnoware is called predictive analytics. The goal of this project is to use a variety of patient characteristics, like their age, and some information about their sleep apnea and diagnosis, as well as titration numbers. We built a predictive model that can tell physicians how these patients are going to be using their CPAP devices when they are prescribed a CPAP machine.

What we found in our research and in our data analysis is that our ability to do these kinds of predictions using this kind of information is very powerful. We can explain almost 70-80% of the variability in CPAP usage by just using variables or characteristics, like age and AHI score, and titration AHI and some initial information on their CPAP usage.

We are at a point where we are taking this model and incorporating it into the product. You will see the output of this model on our physician dashboard, as well as on the individual patient dashboard, as a part of our Care Management Module. Because of that, physicians can now understand how a patient would use their CPAP device over a longer period.

One of the most important things for the industry is to understand Medicare compliance. Medicare compliance provides physicians with the ability to understand how the therapy is being used. It also enables DME companies, manufacturers, and others to get paid for the devices that they are providing to the patient.

With our predictive analytics, we can estimate the likelihood of Medicare compliance for a new patient right at the beginning of their therapy process. As we get more and more usage data for a patient, we can update the Medicare compliance probability until a week or two weeks into the process. This is a very powerful model, it is an exciting feature that is sure to be useful both to physicians, as well as everybody else in the industry and we are extremely excited about it.

Share

Related Articles