In 2002, AME Africa was awarded the contract for the design, implementation and operation of an Electronic Patient Record system at the first world class fully electronic hospital, Inkosi Albert Luthuli Central Hospital in South Africa.

Adopting the well accepted views within the Business Intelligence marketplace, AME Africa undertook a product comparison, and procured an OLAP 'Cube' based product, which is commonly used in the Health industry.

1. The concept of updating the 'cubes' on a monthly basis were not meaningful as the     results were too late for decision making process;

2. The fundamentals of data is such that, defining 'fixed dimensions' is not practical when     more in depth analysis and decisions are required;

Understanding this dynamics led AME Africa to go back to the 'drawing board' and considerable effort was spend with the specialists in trying to 'redesign / rework' the existing environment.

There were improvements in addressing day to day standard analysis but the user could not get to the next level for more real time, complex analysis. Users were not able to analyze the data as and when needed without having some work done on the 'cubes'.

Finally acknowledging that the current technology not capable of delivering solution to our requirements, with started to look out for alternative solution.

Having reviewed many of the 'top end cube' based systems, which still had the same limitations, AME Africa was introduced to the Speedminer DW & BPM. Being very sceptical to the claims of Speedminer, a pilot was undertaken and the outcome of this pilot which addressed the complex issues was astounding, as reported by both the technical team and the end users, in that both of the above challenges were met without any loss of the existing functionality. Furthermore Speedminer, enabled the technical team to 'customise' the user interface to their exact requirements, and hence add to their satisfaction. Then came the real bonus; all of these were being achieved in a shorter implementation period than the previous technology and products. The technical staff did not want to return to the previous method of development, as they were now able to develop quicker and provide the end user quality of work which they were very proud of. Development time of the more complex statistical reports was cut from an average of four months per module to only a couple of weeks. This is a considerable achievement, considering that the base health data had to be transformed and clinically verified, both for quality and appropriateness' before being released to the end users.

Another startling point is that this improvement was seen after only four weeks of on site training. The speed at which the learning curve was scaled can be attributed to the:

1. Flexibility of Speedminer

2. Quality of the support, both from a technical and application
    aspect, provided by the Speedminer team.

Many products reviewed require Dashboard views to fit a predefined look and feel, whereas the Speed Visualizer's dashboard development is limited only by one's imagination. The large array of embedded components (features) of Speedminer, is easily combined to create meaningful views of the data.

Although initially being very cautious and apprehensive, the development team and end users are now very excited about the way of working and the fact that they are making such good progress.

No doubt that, when one considers investing in Business Intelligence solutions, following 'tried and tested' products and technologies may not yield the most effective and appropriate solution. Speedminer DW & BPM certainly should be considered alongside the 'bigger and more established' players in the Business Intelligence market.

"A limiting factor experienced with the cube based model, implemented previously, was that only 16 dimensions could be used to slice the data. Even using half this number resulted in a degradation of performance. With Speedminer this limitation is no longer a consideration. Further, keeping track of changes made to these dimensions was a never ending struggle. As Speedminer builds the "dimension" directly from the transactional records this is no longer an issue."