MTAnalytics 
At MTAnalyics we focus on increasing statistical power to improve the detection of weak association signals between variables. By contrast to data mining, we are using statistical tools that allow us to provide power and p-values in our analysis. In other words, we are specific alike classic statistics and sensitive as data mining.
For clinical trials, we significantly reduce the sample size while keeping the same power. Fundamentally, you will be able to decrease your sample size by 30% while keeping the same p-value, the same statistical power. Obvious main consequences: gain of time, of money, resources etc…
Our work addresses these challenges and aims at providing rationale to design and perform statistical analysis when large amount of data are available. We are not competing with the companies allowing to hire patients faster. We come into play before the design of the trial even started to allow for the design of a trial to include a reduced sample size. We focus in particular on increasing statistical power in order to reduce sample size or improving the detection of weak association between variables by leveraging all available information.
In big data, we provide insights into data sets, predict outcomes, increase the strengths of the models used to explain an outcome based on the analysis of a dataset. Again, the basis of our approach is to improve the detection of weak signals.
Any classical approach is able to detect a strong signal, but when an outcome is not explained (or associated) to one major or a few big variables but by a multiple of weak signals, you need a model to link and find these weak signals, in order to explain a maximum of your outcome.
With our approach, we are able to explain significantly more of the outcome, by reinforcing the models, by finding weak associations hidden in the data sets and missed by the other approaches. In other words, we detect associations that others cannot.
