Medici Technologies was established to support its clients’ data analysis capabilities and to leverage client’s internal data analysis capabilities by providing critical resources and many years of experience in team work and project coordination. When hiring Medici, you are obtaining a structured approach supported by a team of individuals with diverse expertise, an established program management process, and a modular toolbox for rapid data processing and algorithm development.
Via a structured, vetted, and highly visible process, Medici provides services that answer the following needs:
- Due diligence: Do the data results support the conclusions and investment?
- Algorithm validation: Can the results be independently verified?
- Product development: Can we take the data you have and create the product you want?
- Product improvement: Can the product be improved through better data analytics?
Algorithm Due Diligence
Prior to investing in a given opportunity or making a recommendation of investment, the client may want an independent opinion on the validity and performance of the claims. In the area of complex modeling strategies, outright fraud is uncommon but lack of statistical rigor, overfitting the data, unintentional data leakage, and/or unsupported conclusions can lead to overoptimistic claims. In these situations where time and monetary pressures are high and investment dollars large, a second independent evaluation represents a best practices scenario. Medici Technologies can audit or independently reproduce conclusions drawn from existing data to complement the financial and intellectual property audits typically completed prior to investment.
Prior to launching a new product or implementing a clinical trial, an additional level of validation may be desired to protect against a catastrophic result. Given the complexities of algorithm development, an internal group may lack the tools to robustly estimate prospective performance or even unknowingly overestimate performance due to data leakage or undetected errors. Medici Technologies can re-create the algorithm methodologies using independently developed and tested code. The Medici Rapid Assessment System is a modular development environment that enables fast and easy implementation of any modeling “recipe”. Any methods not already in the toolbox can be easily plugged into the statistically rigorous framework. This type of independent algorithm validation can provide the confidence needed prior to initiating a major event.
Medici Technologies can enhance and accelerate a product development cycle. Medici’s goal is to take the data you have and help realize the product you want. Medici provides an array of services to further this goal, including: assisting internal capabilities that may be bandwidth limited with parallel idea exploration. This support leverages Medici’s existing infrastructure of code and processing hardware, and is integrated with project management, risk mitigation and timeline coordination.
Medici technologies represents far more than simply adding additional bodies to the data analysis process. Although Medici technology scientists represent extraordinary domain space knowledge, they operate with a very sophisticated analytical toolbox. This toolbox represents a significant prior investment in processing code development, data flow infrastructure, and results tracking.
Through the use of the Rapid Assessment System, Medici can rapidly process existing data sets using different methodologies within a modular and parallel framework. Key factors or variables under consideration are varied or modified within an algorithm “experiment.” These statistically designed changes are referred to as “switches” within the design of experiments approach. The resulting “processing recipes” contain different combinations of preprocessing techniques, feature selection methods, noise filtering, regression methods and sub-modeling. These algorithm experiments may have millions of possible combinations, but Medici’s efficient methodology can evaluate hundreds to thousands of different processing recipes via a statistically designed search method. The output of the search can be used to identify subtleties within the data as well as identify combinational influences that may be unobserved in single factorial experiments. The resulting statistical analysis of the results enables Medici Technologies to define what parameter switches are most important, how these parameters switches interact, and which switches are detrimental to prediction accuracy. These parameters can then be used for subsequent optimization with high accuracy. The resulting algorithm development process is thorough, highly structured, and very efficient.
The product development process can elicit incredible creativity resulting in a long list of ideas and concepts for solving a given data analytics problem. However, time and internal resources limitations require selecting only a few to investigate, potentially leaving significant performance and understanding on the table. The Rapid Assessment System enables exploration of these valuable ideas in a quick manner with the ability to directly compare performance metrics at different stages of processing. In fact, dashboard comparison metrics can be established for easy visualization.
In summary, the combination of project management, data analysis expertise and the associated tools for data analysis allows Medici Technologies to help your company create the predictive algorithms needed to move your company to the next stage.
Following initial completion of predictive algorithm developments, there is often enormous appetite for improving overall product performance. Such product performance activities can be difficult to pursue due to resource limitations imposed by the start of a clinical trial, completion of product development or product launch. Manufacturing issues can also fully consume internal data analysis resources. Medici Technologies can be used to explore existing concepts, create and examine new approaches and continue to upgrade algorithms as additional data comes available. The messaging capacities of the Rapid Assessment System enable direct comparison of interim and final processing results with historical benchmarks. The capability of providing real time feedback on potential product improvements based upon the most current data available can be remarkably viable in terms of identifying potential problems with the existing product as well as concurrently looking for opportunities for product improvement. Product enhancement areas particularly amenable to post product launch evaluation are improving the number of reported results, decreasing measurement time, decreasing instrument startup time, and ensuring that the product works on a broader demographic group.