Pharmaceutical and biotech companies continue to bet big on AI applications across compound discovery, drug repurposing, and real-time analytics for patient-centric trial design and recruitment. The effective utilization of data and application of AI technologies for gaining insights and decision support will impact the complete value chain for drug discovery and development. Over the next five years, the industry can expect to cumulatively save as much as $50 billion with investments in AI-based products and solutions. This presentation will focus on the commercial aspects and call out unmet needs along with current & emerging applications of AI/ML within the sector. Benchmarking of the AI-enabled competencies of tech vendors will be followed by a brief assessment of the recent trends, growth opportunities, and strategic imperatives. An overview of the emerging ecosystem, partnerships, and selective case studies will be discussed. Key insights will be shared including: 1) themes driving AI application across the drug discovery & development value chain; 2) dynamics around regional trends, vendor eco-system, innovative business models, and game-changing companies, and 3) investment opportunities and growth potential.

AI/ML in drug discovery has recently garnered hype and funding, with exciting academia and industry results. Yet, whether we have seen transformative value-generating breakthroughs is still a topic of debate. What is becoming clearer, is that AI/ML can make existing drug discovery processes better, more efficient, and also provide previously elusive insights. Here, we show how AI/ML, a key pillar in the Frontier platform, enables drugging the undruggable.

Finding druggable targets for complex chronic diseases is particularly difficult using existing drug discovery approaches. This leaves significant pockets of patients with unmet medical needs, but also many opportunities to reposition active molecules originally designed to treat another disease, which acts on mechanisms relevant to the needs of currently ineffectively treated patients. High-resolution patient stratification based on combinatorial analytics offers an accurate and scalable route to map effective drugs to patient subgroups across multiple new disease indications.

Protein-based therapeutics now make up approximately seventy percent of the pharmaceutical market and have made remarkable impacts on the lives of people affected by numerous different types of diseases. However, the field has been limited by what is observable in nature. Machine learning and artificial intelligence are helping to change that by elucidating the genetic code underlying the function of proteins and developing a generalizable set of rules and principles by which proteins are created and function. By combining machine learning with wet-lab research, Generate Biomedicines’ technology platform enables de novo antibodies, peptides, enzymes, cytokines, and other previously undiscoverable protein therapeutics to be rapidly developed at unprecedented speed and scale, potentially reducing the cost and time associated with therapeutic development. The speaker will discuss the important role of machine learning in the future of drug development and how Generate’s approach to generative biology could unlock the key to the medicines of the future. 

Salve Therapeutics is developing a bioengineering software platform called VirCAD (Virus Computer-Aided Design) that will: 1) cost-effectively mine the human virome (and others) for new therapeutic modalities that will enable exploration of this therapeutic space; 2) discover novel drug candidates via the design, modeling, and simulation of novel viral particles in silico; 3) manufacture better viral delivery methods that would expand the available viral biologic tool kit to produce better precision medicines, and 4) help treat and potentially cure the many inherited and acquired.

In any antibody drug development campaign, there is an emphasis on identifying clones that can deliver high product titers at the desired quality. For the selection of highly productive, stable, and robust monoclonal cell lines able to serve both clinical phases and the market, a large number of clones have to be screened, filtered, and further processed, making the clone selection process very demanding in terms of resources and time. The recent advancement in AI and ML technologies has created the opportunity to predict the outcome of bioprocessing, promising a pipeline with widely applicable attributes. In this project, high-dimensional multi-omics data, derived from 1000 cell lines producing 11 different antibody formats, were collected while in a very early cultivation step of cell line development. The data library created, follows an effective strategy for annotation, storage, quality control, and data harmonization, to simplify a rapid data access framework suitable for multimodal computations. The final collection is utilized to develop ML methods for identifying early predictive marker networks by screening clonal profiles across various production batches in the early stage and analyzing their correlation to late-stage performance. Our approach aspires to transform cell line development into an in silico-based methodology, offering the benefits of lower experimental effort, process transparency, clear rationality behind decisions, and increased process robustness.

At Biolexis, patients come first. Our MolecuLern platform is capable of developing new medicines for these patients at a fraction of the time and cost. MolecuLern is a proprietary, real/empirical deep learning protein hot-spot screening technology, with innovative human-in the loop learning workflows for predicting 40+ molecular properties. MolecuLern is uniquely suited for identifying and designing novel drug-like compositions of matter, by simultaneously prioritizing compounds based on selectivity, off-target profiles, and developable criteria, all with unprecedented speed and accuracy. The audience will learn how our MolecuLern technology is capable of developing novel small molecules, and how we differentiate ourselves from other AI/ML platforms.

Selecting the right patients is a key aspect of drug development. This talk will discuss the use of AI and machine learning in selection of right patients, including natural language processing, structured query, classification, ranking algorithms, etc.

BioSymetrics uses a phenomics-driven, AI-powered approach to the discovery of novel targets in cardiovascular, neurological, and rare diseases. Our automated experimental platform validates target predictions in vivo using zebrafish and custom computer vision software that rapidly quantifies results. Learn modern rethinking of phenomic-driven target discovery that incorporates clinical data, public data, and experimental data to build translation up front. We will share platform technologies as well as applications to pipeline programs.

In this work we describe a process for automatically expanding chemical space around a hit using validated multi-step chemistry and readily accessible building blocks. This multi-dimensional extension upon traditional one-dimensional library enumeration defines a large, but still computationally manageable search space to identify analogs with improved molecular properties. Using machine learning models trained on historical activity and ADME data, we can rapidly and efficiently explore local, accessible chemical space.