Especially with respect to animal testing, scientists are making a great effort to reduce experimentation on animals, something that is increasingly rejected by society but which is still necessary; however, it seems that there is a possibility, recently tools have been implemented that promise to reduce not only the time of development of drugs but also the number of tests on animals, which means greater economic benefit by putting a drug on the market in less time.
Image edited by @emiliomoron, original sources: 1, 2.
For artificial intelligence, as I mentioned in the title of this post, is helping to shorten these figures and improve the success rate. As we already know, through AI machines can learn, either through processes such as automated learning or deep learning, terms that we hear more and more often.
AI has been used in medical chemistry for a decade, employing computational techniques at all stages of the development of some drugs, from the identification of the cause of the disease to the selection of the candidate drug molecule that will have a therapeutic action on the disease. They also include applications such as modelling the process of absorption of the new chemical by the organism and can even be applied to select the ideal patients for clinical trials.
By developing artificial neural networks, machine learning is being used to predict any biological property of a molecule without the need to synthesize it in the laboratory or also to predict its effect on the organism without having to resort to animal testing. The machines learn from all the data that are loaded by the scientists, such as chemical structures and properties already known, such as physical-chemical, toxicological or pharmacological parameters, which they want to predict in new molecules.
Image edited by @emiliomoron, original sources: 1, 2, 3.
Already in January this year we saw proof of this, the BBC released a molecule developed entirely by an artificial intelligence that would enter its testing phase in humans, this is the molecule DSP-1181 which has been developed as a treatment for obsessive-compulsive disorder (OCD). Mathematical algorithms were used to design millions of potential chemical structures and then filter them to select which molecule to test.
With the current health crisis caused by covid-19, for which there is no effective treatment, AI has the full potential to develop new treatments and slow the progression of the virus. For example, models can be used that use the database of drugs already developed, using the structure of the antivirals already used against the family of coronaviruses, both those that have been successful and those that have not, with all this information, the AI will learn, and could propose the chemical structure for a new molecule that has activity against the virus, saving useless testing and the corresponding loss of time.
Saving time and economic resources are crucial factors in a crisis situation like the one we are currently facing, and AI may be the key to identifying antiviral drugs for this or future pandemics.
Thank you for reading me. Until the next post!