Aria Pharmaceutical's Implemented AI for Generating New Innovative Insights

Radin, for example, took the audience through the discovery process for TXR-711, Aria's flagship drug for treating systemic lupus erythematosus (lupus).

Fremont, CA: When compared to standard ways, Aria Pharmaceuticals' innovative computational approach to drug development has raised its success rate by 30-fold. At this preclinical stage, small molecule firm, this enhancement was critical in developing the 18-compound pipeline.

According to a statement by Aria CEO Andrew A. Radin, BioSpace that the 30-fold number is based on comparing in vivo milestones in animal efficacy trials. Companies generally test 100 compounds and anticipate one or two to demonstrate disease-modifying signals using existing methods. They observed three out of 10 with their approach. It's a significant increase, especially considering Aria's focus on creating therapies for complicated and difficult-to-treat illnesses with poorly known underlying biology. It cuts medication development time in half.

Radin also outlined the technique in an early January presentation at Biotech Showcase. He mentioned that they don't get fresh insights from the literature, and instead, they employed computer modeling to generate new insights. They discover chemistry paired with a novel theory as to why that chemistry should be safe and effective. More importantly, they develop and test several sickness hypotheses at once, instead of just one at a time.

Radin, for example, took the audience through the discovery process for TXR-711, Aria's flagship drug for treating systemic lupus erythematosus (lupus).

The first two processes necessitate the use of a computer. Meanwhile, to begin, Aria's artificial intelligence (AI) engine uses data from around 25 sources, 65 methodologies, and a chemistry library of over 2 million molecules to identify essential biological aspects of the illness and link them to molecules with known chemistry. The AI engine selects those with the highest expected efficacy based on building an in silico model of the condition.

That model takes into account gene expression, microRNA expression, protein expression, clinical and phenotypic data, and other factors, including pharmacovigilance data from approved medications from the US Food and Drug Administration (FDA). Around 3,000 chemicals have been identified as being linked to lupus.

Radin also explained that, after that, they screened the hits for unique modes of action, safety profiles, absorption, distribution, metabolism, and elimination (ADME). It brought the total number of compounds down to 80.

Human perceptions are crucial at this moment. A thorough dive led by a Ph.D. evaluates the relevance and safety of the identified mechanisms of action, as well as the anticipated intellectual property path.

Radin also estimates that the three processes – from the start to the selection of pathways for preclinical models – will take roughly four weeks. Aria collaborates with contract research groups to evaluate the various modes of action and in vivo efficacy in the following phase, preclinical evaluation.

TXR-711 enhanced kidney function and decreased inflammation in preclinical animal models, demonstrating that its effectiveness is equivalent to cyclophosphamide – a medicine used to treat severe flare-ups – but without the drug's severe side effects.