|Table of Contents|
|Epilepsy Monitor on HCP – Introduction|
|Epilepsy Monitor on HCP – Patient Mobile App|
|Epilepsy Monitor on HCP – Family and Relatives Web App|
|Epilepsy Monitor on HCP – Doctors and hospitals SAPUI5 App|
|Epilepsy Monitor on HCP – Predicting possible seizures by using PAL|
|Epilepsy Monitor on HCP – Conclusion
You can find an introduction of the story at Epilepsy Monitor on HCP – Introduction
In this part we’ll explain the most exotic part of the application. The part that shows the power of HANA integrated in HCP. The part that can help healthcare.
In the mobile app we show two results of the prediction. On the first screen of the mobile app we show the percentage of a possible seizure in the next five minutes.
On the last screen we show the trend of the heartrate with the heart rates in the next five minutes. (red line)
So now’s the question, what’s behind all this. Thanks to PAL integrated in HANA you don’t have to be an analytical expert. You can just use the predefined formulas like you can use function modules in ABAP 🙂 Therefore we’ve created our own Stored procedures that will call these PAL functions and return the result.
In the blog about the mobile app, Epilepsy Monitor on HCP – Patient Mobile App, I already mentioned that we use a Java Servlet to communicate with HANA from the mobile app. In this Java Servlet we call the stored procedures in HANA.
For the web ( Epilepsy Monitor on HCP – Doctors and hospitals SAPUI5 App ) and sapui5 ( Epilepsy Monitor on HCP – Doctors and hospitals SAPUI5 App ) app we can just use the XSEngine to call the Stored Procedure.
So that’s how the mobile app and web apps call the Stored Procedure in HANA. But which PAL functions are we using in HANA for our predictions?
We used two functions:
1. Double exponential smoothing model
The first one will analyze all the historical heart rates of the patient, look for a trend in the heart rates over time and predict the next heart rates.
2. C4.5 tree model
This model is used to decide if the predicted heart rates are possible seizures. Therefore we first use the previous model and with that result we use the “C4.5 tree model”. This will use all the data from the HANA database like all patients, seizures, heartrates, questions&answers to decide if a heart rate is a possible seizure. It will create a complex decision table from all the different data.
For the red line in the graph we use the first PAL model:
For the possible seizure risk on the first screen of the mobile app, both models are used. First the “Double exponential smoothing model” followed by “C4.5 tree model” to decide if it’s a possible seizure.
That’s how we’ve used the PAL library of HANA. Thanks to this library you don’t have to be a scientist to analyze and predict data.
Check our full demo on youtube: Epilepsy Monitor on HANA Cloud Platform Demo – YouTube
Next part of our Epilepsy Monitor: Epilepsy Monitor on HCP – Conclusion