Throughout the history of the pharmaceutical industry development, drug discovery was, for the most part, a serendipity-driven tedious process of "trial-and-error", with poorly predictable expectations and extremely low success rates of bringing initial drug ideas to the market. This was conditioned by quite limited understanding of biology, lack of experimental data, and the absence of modeling tools and computational resources. The situation started to change for the better in the 1990s with the emergence of so-called rational drug design,
pioneered by such companies utilizing structure-based computational discovery strategies.
In about the same period, the progress in a number of research-enabling technologies started to unfold rapidly, including high throughput screening, combinatorial chemistry, high throughput sequencing, and omics analytical methods — all generating massive amounts of data in the biomedical field. Simultaneously, the world has been witnessing the unfolding revolution in hardware development (high-performance computing, GPUs, and data centers of the new generation), and computational data modeling methods, including the advent of Artificial intelligence technologies.
In the early 2020s, a qualitative leap has been made in the area of Deep Learning and the underlying modeling architectures — deep neural networks. The Famous ImageNet competition in 2012 marked the rise of convolutional neural networks in prominence, and in 2014 Ian Goodfellow introduced Generative Adversarial Networks
, which would later become one of the most advanced methods in generative chemistry and chemical drug design, owing to our works, and works by some of our peers.
Around the time deep learning was rising in prominence in such applications as image recognition, Insilico Medicine was founded in 2014 as the company pioneering the application of deep learning technologies for drug discovery. Those were the early days the pharmaceutical artificial intelligence and Insilico Medicine had to perform dozens of complicated pilot studies with the pharmaceutical and biotechnology companies to learn where AI can help accelerate and cost-down the process, or augment human decision making leading to the development of the creation of the first end-to-end AI-driven drug discovery platform.
The company's AI-first strategy, adopted from day zero, and a laser-sharp focus on building, validating and iteratively improving its deep learning-based generative biology and chemistry pipelines of models gradually led to the emergence of what we call "end-to-end" platform, Pharma.AI, which covers a whole spectrum of drug discovery, starting with disease modeling and target discovery — to hit discovery and various stages of preclinical optimization. The platform also includes InClinico — a specialized system for predicting outcomes of clinical trials. Today, the accumulated experience of Insilico Medicine in the application of advanced technologies to biological and chemical modeling, and possession of the most comprehensive drug discovery platform on the market, gives the company a strategic advantage in the race for discovering new first-in-class targets and molecules in record times and at low budget.