AutoDiscovery now supports the consolidation of up to 50,000 records and 90 measures (variables).
Handling larger datasets implies being much more efficient in the data consolidation, visualization and exportation so these processes have been also boosted with optimized algorithms.
Much more to discover!
Although correlations are very useful statistical calculations to identify potential cause-effect relationships, they are not good enough to assess the impact of qualitative variables on numerical measurements.
Beside correlations, AutoDiscovery now also evaluates ANOVAs between every qualitative variable and every numerical variable in each subset of the consolidated dataset (complex variables such as ratios are not considered).
in addition to ANOVAs, AutoDiscovery now assesses how time impacts on every numerical variable determining whether any of them tend to systematically increase/decrease in some particular conditions.
The discovery process may now be focusedto a subset of your variables so that the time required to analyze their relationships is drastically shortened.
That subset of variables is called "variables to explain", which reflects the underlying goal of AutoDiscovery: letting the data tell the story.
Much more exploration!
It appears reasonable to think that the larger the dataset evaluated, the higher the number of revelant relationships to be found and therefore the harder to reach to sound conclusions.
That's empirically true. We've tested it with several real clinical datasets and projects and reached the conclusion that an advanced tool to explore specific parts of the new body of knowledge generated is a must.
The Hypo Booster tool enhances the definition of hypothesis as an advanced filter to the exploration process.
The idea is quite simple: express your exploratory hypothesis in terms of relationships between the variables, specify the conditions to focus on and AutoDiscovery will find out only the correlations and factors involving these subset of variables.
Not only our engineers were obsessed over providing better tools to handle such amount of information but also over enabling researchers to do it faster. At least faster than the previous version of the software.
And they succeeded!
Thanks to a streamlined architectural design and multi-threading algorithms, they finally improved the efficiency in the handling of data, statistical calculation and user experience in general to the point that AutoDiscovery multiplied its speed by 4.
If you're a new user, signup with your email address, password and first name. You'll receive a confirmation email to activate your license and get access to the setup tool.
3. Install the software in your computer following the setup wizard.
The majority of the improvements described here were the result of excellent ideas combined with a close cooperation with brilliant professional research teams in hospitals and public labs around the country.