Dan Dascalescu's Homepage
/essays/medical singularity

Reaching the medical singularity

paraphrasing a speculative idea by Evgeni Duhovny


Imagine an input in an initial state, a desired state for that input, and some environmental variables (matter and energy). You have to bring the input from the initial state into the final state by applying a series of rules to input and the environmental variables. For example, if the input is water and the desired state is ice, the solution is to lower the temperature below zero degrees Celsius.

Let's go one step further. Imagine the input is a chemical substance, the desired state is another chemical substance, and the environment variables are additional chemical substances that you may use, and physical factors (temperature, pressure etc.). You want to determine how to transform the input substance into the desired substance. In many common cases, there are already chemical formulas that are guaranteed to work. In other cases, you can go through a series of chain reactions and reach the desired state. These chain reactions can perhaps be inferred using methods of analytical chemistry, or even calculated by a computer algorithm based on existing knowledge, trial, and error.

Let's go one step further. The input is a culture of cells infected with a bacterium, the desired state is those cells without the infection, and the environment variables are the same as in the basic chemistry problem above. By synthesizing a particular antibiotic, you can eliminate the bacterium, in effect curing the cells. The cure for the cells can be determined based on methods of biochemistry, or, again, calculated using a computer algorithm and (more) existing knowledge, trial, and error.

Let's go one step further. The input is a living organism with a particular disease. The desired state is the living organism without the condition. As long as the input and the desired state can be modeled (digitized into a computer), the cure can theoretically be calculated using a computer algorithm. In practice, factors that come into play are the resolution of the models and the complexity of the algorithm.

Let's go on step further. The input is a human with a particular condition - aging. The desired state is the same human at age 25. This article attempts to prove that the solution can be calculated using a computer algorithm.




Utopian? Today, yes.

What is the status of this document?

First draft. Very little scientific research done. Numbers may be off by orders of magnitude.

Living organisms are very complex and modeling them far exceeds computational capabilities

A1: In many cases, we don't need to model the entire organism, but only key parameters. This reduces the search space significantly.

A2: computational capabilities have followed Moore's law: microprocessor speed, storage price and supercomputer power have all been improving exponentially. Moreover, technological paradigm shifts keep happening. The future of computing looks promising with optical processing and quantum computing, the latter opening the door to much faster processing than any classical computer.

A3: living organisms have many mechanisms of homeostasis (self-correction). This enables many slightly different solutions to work, increasing the solution space.

What if you do end up having to model the entire organism?

A: The resolution of noninvasive brain scanning has been increasing exponentially as well.

Isn't it simpler to inject in the body an army of nanobots and have them fix things individually?

Perhaps. But it may prove that nanobots come with additional problems: too large compared to cell size, can't be eliminated nor reprogrammed once they performed their mission, limited range of function (e.g. can't significantly influence cell metabolism)