Case study card:
Liver disease diagnostics
The client wanted a predictive tool to improve liver disease diagnostics.
There are many elements that combine to produce a disease.
These are factors such as lifestyle and diet, as well as proteomic or genomic elements.
Usually, efforts to combine data sources do not work very well.
We developed a multimodal AI application (one that combines all factors into an overall analysis) to predict the presence of liver disease from a multitude of data verticals.
Typically, this kind of approach fails due to a lack of compatibility between data sources, so our proprietary system creates self-organising software units to ensure comparability.
This supplier has a proprietary system (patented, funded) consisting of agents that self-organise into abstract software units that can be integrated with related algorithmic models.
This is unlike deep learning networks, which often lose semantic information with accuracy (i.e. what and why?), instead merging many smaller (less complicated) models with a software (not algorithmic) approach. This robustness and explainability has been welcomed by medical practitioners.
In simple terms, their system can merge siloed models and data sources to create all-encompassing predictive capability.