Interpretated artificial intelligence to discover drugs

Over the past decade, Discovery of drugs with artificial intelligence It appeared as a transformative tool to accelerate the discovery of pharmaceutical medications by automating tasks and providing predictive visions. It can be used to determine goals, vehicle design, examination of molecules, implementation of the reverse artificial planning, forecasting of the drug, absorbing, distributing, metabolism, secretion, and pharmaceutical kinetic properties, thus shortening the timelines for the research significantly (1).

However, the success of the development of drugs supported by artificial intelligence (AIDD) still depends on bridging the gap between wet and dry experiences, which requires great experience in life sciences and algorithms. In addition, the accuracy and reliability of predictions by machine learning models depend on the selection of the data set, the model training processes that result in predictive results, and how these predictions agree with the progress of the project.

Creating a revolution in medicine by discovering medications using artificial intelligence

For example, Healx has succeeded in using the obstetric rivalry network to determine and develop a new X fragile syndrome in 18 months. The drug was later developed for clinical trials, which confirms the value of interpretative artificial intelligence. Likewise, Collins and his colleagues (234) used an explanatory DL model to give priority to new antibiotic candidates against staphylococcus resistance to methyline.

To enable precise predictions, automatic learning models should capture the chemical space of the drug molecules using molecular description features (also known as prescriptions). These features are used for encryption Molecules To a fixed length chains that can be stored and processed by computers. It is a prerequisite for mathematical tasks such as virtual examination, vs: admet, complex search, ADME/T prediction, and reverse artificial path planning. However, many machine learning models are not designed taking into account the possibility of interpretation and often behave like black boxes. This shortage of transparency makes it difficult for researchers to understand how the model reaches its conclusions and verifying and improving the model.

https://www.youtube.com/watch?v=oCVry1My07k