Would you like to know if you are going to have a heart attack? Would you like to know what is the risk for you?
Cardiovascular disease is a huge problem all over the world. 610,000 people die of heart disease in the United States every year and one of my family members passed away because of the heart-attack. So we thought that we could build an application which estimates the ratio of heart-attack risks.So we can prevent them by managing or life.
What it does
User registers the application and fills the necessary data for the app such as of smoking per day, how many years the user have been smoking. Other necessary data is taken from user's Microsoft Band. (weight, height, body-activity etc.)
Based on the recorded real data from hospitals we found, we have set a Machine Learning system to the cloud in this case Microsoft Azure. Based on the information such as weight, height, race, age, smoking, sex, calories burnt, body activity, heart-rate, steps etc. We have trained that Machine Learning with the data provided. It's creating an artificial intelligence on Azure Machine Learning. It's learning. It's producing data.
All those data is taken from Microsoft Band, real-time. Current heart-rate, body activity, calories burnt, steps taken effects the user's ratio uploaded on the machine learning. Also we have created a tile on Microsoft Band for user adds which user can press while smoking. So it adds one more to smoking table of user.
"Heart Care" application is connected to Azure. So, when the user hits the calculate my risk button, it is basically going to machine learning on Azure and fetching the data to the Heart Care.
It's a learning system. As much as we provide real data of people, the Heart Rate becomes more precise. Theoretically, with the correct data provided, we can have %99 precise result.
How I built it
First we have collected real data from a hospital. We trained the data on Microsoft Azure Machine Learning. We fetched the real data of user from Microsoft Band. We have created a Windows Phone application which communicates with all of them.
User's data gathered data goes to ML on the Heart-Rate application. ML compares the data and returns a ratio. Best side of it, it is learning and getting better at the same time.
Challenges I ran into
Learning a new thing can be challenging always. Machine learning was sort of challenging.
Accomplishments that I'm proud of
Heart Care application has a social cause. We are really happy to help people. With our application people will be able to maintain their life and they avoid the heart attack. We are really proud to have a change the technology to help people.
If we can help one single person, that is the biggest prize for us.
What I learned
We have discovered the Power of Azure's Machine Learning. It's very impressing.
What's next for Heart Care
We really want the Heart Care helps people. Regarding to that, we are going to publish it to the market for free.
What is next for Heart-Care; we are going to be using Microsoft Band 2 and we are going to be able to fetch the data of a person exactly how much stressed he or she is. It's called Galvanic Skin Response(GSR). With the GSR we will also add it to Azure Machine Learning.
So even the how stressed you are at the moment, may be able to affect the values. We believe that Heart Rate will become so popular that even will be used on hospitals.