Computers and A.I. have shown that they are great at play games. So turn the problem into a game.

Level 1: Experiment and deduce.

Science is essentially a game of 20 questions or guess who.

Player 1 holds a secret.
Player 2 must guess the secret, by probing
Player 1 homes in with eliminating questions and after a number of tries makes a guess. 

In science, it is the system that holds the secret, and it is the job of the scientist to work out the secret. A scientist probes using experiments.

It is not only unravelling an information, mapping. It is also unravelling rectification, cure.

Both of these can be thought of as secrets in the Codebreaker's Conundrum. "Find the (unknown) cure" and "Map the (unknown) area"

A simple version of the game is chemistry and a chemist that seeks to determine the makeup of an unknown substance. A chemist performs a series of tests and each result provides a clue to the element or molecule.

The method of deduction is one or more, for instance, comparison, set patterns, maximal elimination, predetermined models, and so on.

In one game, it can determine any unknown element depending on the outcome of a set of experiments. These are set pieces, not even A.I. the machine could perform this role without any human intervention.

Level 2: Riddles

In this game two images are taken, the disease state and an optimal health state, the difference is the cure. So if the health state is 7 and the disease state is 4 then the cure is 3.

Level 3: Equalize health.

In this game two images are taken, the disease state and an optimal health state, the difference is the cure. So if the health state is 7 and the disease state is 4 then the cure is 3.

The researcher must conduct experiments to manufacture the 3 and also use an existing knowledge base that could help.
The manufacture of 3 is an iterative process.

The process involves imaging technology, analysis and comparison, perform experiments to determine a cure.

With A.I. tokenisation, the association with numerics is relevant.