The San Juan Islands Archipelago in Washington State.

This web site is about two major applications of the system – medical diagnosis and the generation of training data for neural networks.   The Bayesian system is an important component of these applications.  However, detailed information about it, especially its internals, are highly proprietary and treated as trade secrets.  We can understand the curiosity you might have about the system but will not respond to requests for detailed information about it.

Conditioning by Conditional Probability Tables (CPT) for each combination of conditioning node states would, of course, be impossible because of the number of CPTs required.  However, for conditioned nodes defined as “Noisy Max”, a CPT is required for only the first n – 1 states of each conditioning node, where n is the number of states of a conditioning node.  Although the “Noisy Max” simplifies the specification of the conditioning, evaluating a node defined this way can still require a non-linear evaluation time.  For the two commercially available systems we have tested, this is the case.  However, we can evaluate in linear time.  For example, if we had 10,000 conditioning nodes instead of 5,000, it would take twice as long to evaluate.  How we do this is proprietary.

For the world, it would benefit for the reasons given – avoiding the costs, both financial and human, of a new pandemic and the potential for improving global heath.  Since the acquisition would also place the Bayesian technology in the public domain, the world could also benefit by other applications of that technology beyond medicine.  A billion is definitely a bargain as a purchase price.

For us, we would lose the exclusive opportunity to develop and market applications of the technology.  For example, we could develop a service to provide medical diagnoses.  It’s not often that a service could be of value to everyone on the planet at some point in their lives.  Experiments in running such a system in the Cloud have shown that a complete diagnosis (from initial presentation to final diagnosis) can be performed for a few cents (and in the future, even less).  This means the profit margin could be close to 100%.  There would be many variables involved in calculating an optimal price for a diagnosis but estimates we have made show that annual profits could be in the low billions.  And that is only for medical diagnosis.  A billion is definitely a bargain as a selling price.