On April 27, 2020, AI & Medicine, a company that provides a unique artificial intelligence (AI) drug research and development platform, announced that it offers solutions and expertise on ADMET prediction for pharmaceutical enterprises and medical institutions across the globe.

ADMET refers to drug absorption, distribution, metabolism, excretion and toxicity. The early assessments of ADMET properties undoubtedly play an important role in small molecule design and screening in order to avoid costly and unnecessary failures, but this is no easy task. The increasing participation of AI in this field in recent years has strengthened the ability to predict ADMET properties quickly and reliably, which largely helps to exclude compounds with potential ADME issues in the early stage.

The accuracy of AI-based ADMET property prediction is achieved based on Big Data, Machine Learning and Deep Learning technologies. Below is short description of these prediction methods and procedures.

Data Collection
All the obtained data are collected from peer-reviewed publications, public databases through manually filtering and processing.

Data Set Preparing
Pretreatments is a necessary step to guarantee the quality and reliability of data. Compounds without explicit description for ADME/T properties will be removed. Only the unique ones will be reserved.

Descriptor Calculation
Molecular descriptors are applied to further model building, which include constitution, topology, connectivity, E-state, Kappa, basak, burden, autocorrelation, charge, property, MOE-type descriptors, fingerprints, etc.

Descriptor Selection
Some uninformative and interferential descriptors are eliminated and regression models are built using five-fold cross-validation method.

Modeling algorithms
Different modeling algorithms are applied to develop regression or classification models for ADME/T related properties, including RF, SVM, RP, PLS, NB and DT.

Performance evaluation
To evaluate various ADMET properties, a series of high-quality prediction models would be generated and validated, applying fivefold cross-validation and using commonly used parameters.

“AI & Medicine is dedicated to effectively extracting structural features through deep neural network algorithm. We can virtually predict and evaluate ADMET and other properties of small molecule structures on cell, protein and disease level. Our integrated AI-powered drug discovery platform enables our clients to conveniently perform several types of drug-likeness analysis, ADMET endpoints prediction, systematic evaluation and similarity searching. Although drug discovery is risky, lengthy and with high attrition rates, our ADMET prediction model would be of assistance and largely simplify the drug R&D endeavor,” says a spokesperson from AI & Medicine.

To learn more about AI & Medicine’s ADMET prediction solutions, please visit https://aimed.protheragen.com/solutions-for-admet-prediction.html.

About AI & Medicine
AI & Medicine owns an AI-powered drug discovery platform that is capable of offering the best portfolio of medical and scientific solutions in areas like drug R&D, medical translation, medical imaging, medical therapy and research system, and more. With the broad application of AI technology in disease target prediction, high-throughput data analysis and system biology modeling, AI & Medicine is missioned to helping accelerate the drug discovery pace of medical institutions and pharmaceutical enterprises worldwide.