top of page
  • Writer's pictureNikos Tzagkarakis

The next step of a Bayesian Brain | Scientists can now brainstorm with Tzager to find solutions for

Updated: Feb 1, 2023




For the past months we started spreading the world about Intoolab’s milestone of building the closest to a Bayesian Brain, with human-like logic for Healthcare. This breakthrough is a proof that Bayesian Networks are finally here to stay, by being the missing link between Deep Learning and us humans. This leap in our technology allowed us to offer several new improvements on features to our customers, like the Ask Tzager feature (that allows researchers to dive in several layers of causal graphs from Genes to whole Systems), the Workflow feature (that allows scientists to reprogram Tzager’s Bayesian Networks by focusing its attention to specific paths), all the way to Document and Literature intelligent modelling.


Now it is time for the next step, fulfilling the promise of the most advanced human-A.I. interaction in a Tzager’s feature, that we call BrainStorm.

BrainStorm, as its name suggests, is the ability of Tzager to have long back n forth discussion-like interactions with scientists, that allow for an unprecidented path to conclusions. When two humans interact in a brainstorm session together, they basically share a mutual understanding on the causality and the connections of the field and topic that is being discussed, which allows them to guide each other to a mutual conclusion.

This is exactly how Tzager’s BrainStorm feature works.

For the first time, the capacity of an A.I. and the logic of a human are interlinked through a shared understanding of how the world works. This is achieved through Tzager’s Bayesian Networks, that emulate the way we humans understand How the Body works, How Diseases work and How Therapies work.


The way BrainStorm works is very simple.

The user initiates the discussion with a question or objective (e.g. “I want to cure Crohn Disease”, “I want to know why a specific drug works” etc) and the BrainStorm session begins. In the objective “I want to cure Crohn Disease” Tzager understands that the user wants to unveil the therapeutic path of the specific disease, so as a first step it proposes therapeutics and drugs (if the phrasing was different e.g. “What causes Crohn Disease” then Tzager’s response would be different). Then the user can for example ask “Why infliximab?”. Then Tzager will continue in the same train of thought and answer what are some main connections of infliximab and crohn disease in the therapeutic path of the specific disease. Then the user can either continue in deeper questions all the way to specific genes, chemicals, phenomena etc, or she can ask to see a partial causal graph that explains the discussion so far.

This is just part of the most important causal connections that can help the users take the next step in the specific brainstorm session. If the user wants to continue to a different path, she can for example ask “show me the connected proteins”.

And Tzager will give back the most important connections according to the discussion so far.


There is no real limit on how deep or wide the BraiStorm session can be. The idea of this feature is for the scientists to be able to identify connections and causalities in ways that was impossible before. Users can now communicate with an Artificial Intelligence agent, in order to find out how specific connections work, what are some Novelty connections, why specific therapies work or do not work, why disease or symptoms behave the way they do and much more. The path of the discussion depends on how the user asks.


Disclaimer: Tzager is NOT just another chatbot, in fact it’s not a chatbot at all. It will not respond “ Good day” or tell you the weather, because this does not need an actual understanding from the agent, plus these problems can be solved better by just using Deep Learning.

Tzager is not a pre-trained chatbot that tries to predict a specific answer depending on the question without actually understanding anything. The big difference on how Tzager works, is that it understands Healthcare in a similar way to humans by reconstructing the causality of the world based on experience.

For more information you can read our last blog closest to a Bayesian Brain, with human-like logic for Healthcare that described the technology behind Tzager.


We are more than excited to announce a public Beta of the BrainStorm feature where anyone can have up to 5 back n forth sessions with the feature, just to get a taste. All you need to do is go to tzager.com, click the “test BrainStorm” button, give your email (for authentication reasons) and start brainstorming with Tzager!


How can you use Tzager in your projects?

The most important aspect of our Bayesian agent is that it is an API. Currently Tzager is used 1) as an endpoint on existing customer platforms, 2) as a Python library from data scientists and 3) of course through our own platform on tzager.com. If what you are looking for is an intelligence that will give you answers on how concepts are connected, model your solution, give you novelties and causalities from Genes to Systems, there is definitely a solution for your case!

More updates and tests are coming soon from us and our happy partners! Till then, let us know how your BrainStorm session went!

4 views0 comments

Comments


bottom of page