Just to be clear, the app does not reload when you apply a selection. A selection limits the data in your current data set, which means all calculations in the visible visualization are updated according to your selection.
This is usually considered the strength of the product, that you select and see the results. This is a core part of doing discovering what the data can tell you. I have never heard a end-user express this to be a negative thing.
The only scenario where I can imagine this being an issue, is if your data model and visualizations are not optimized for your needs and therefore give you a longer update time. Some times there might be a need for you to force several selections prior to calculating an object, for these scenario look into using "Calculation condition".
Can you please elaborate on what the actual problem is with the selections being applied?
You can not disable the calculation technology. You can mitigate the performance impact.
If you intend to use Direct Discovery (DD) you have to review the setup, and perhaps optimize your DD settings to for example allow for longer cache time on the Qlik Sense server. The idea would be to minimize the need for queries to the data source, as this is where you see the performance degrade.
As I mentioned previously the Calculation Condition on visualizations enables you to define a criteria for when the object will be calculated and rendered. This will practically mean that you will use more limited data sets for calculation, and also have less calculations.
For Big Data solutions I would recommend that you look at On Demand App Generation as a concept of dynamically generating smaller apps for your business users, base don initial selections. This completely removes the DD from your app setup, and gives a much smoother in-memory based user experience.