Sun Pharma-backed AIRA Matrix, a company set up and funded by Sun Pharma founder and managing director Dilip Shanghvi’s family office, is taking a shot at AI-led pharma research and development [R&D]. The company provides artificial intelligence-based solutions for life sciences applications, and has developed deep learning-based products and services to help pharmaceutical companies expedite the discovery and development of new drug molecules by reducing risks, costs and time of at least a couple of key phases in drug development, drug discovery and preclinical trials.
Bringing a new drug to market typically involves multiple phases, including early discovery, preclinical research, clinical development and regulatory approval. The whole process can take 10 years or even more, and cost billions of dollars. But in the end, only one in 10,000 compounds initially considered makes it to the market finally, according to estimates. Hence, there is an urgent need for solutions that speed up go-to-market time and reduce the development cost of drugs and vaccines. A number of technology driven tools have been developed to assist research in the last couple of years.
With 17 papers already published, AIRA Matrix has been selected by GE Edison accelerator. Besides a small grant, it gives the company an opportunity to work with professionals across various disciplines of GE Healthcare and benefit from integrating software products within GE’s ecosystem.
Set up in 2011, AIRA Matrix initially worked on internal product development with a few collaborators to provide feedback and additional feature requirements. The aim was to create a product suite for prostate cancer that not just automates the current error-prone manual reporting processes, but helps in effective treatment decisions throughout— from screening, disease diagnosis, risk stratification and prediction of disease progression and biomarker expression.
“After proof of concept was established from publicly available data, and in the process finished among the top at the Global MICCAI 2019 competition, we forged collaborations with development partners in India, the U.K. and the U.S. Key aspects of the product suite have been developed and are currently in clinical validation,” says Chaith Kondragunta, chief executive officer, AIRA Matrix.
The company’s hybrid deep learning-based models predict potential toxic effects of a new drug molecule under investigation. The models analyse data from multiple modalities to help with the crucial go/no-go decisions in the selection of the safest drug molecule. The risk road map and failure point predictions significantly reduce the time and resources needed in this early phase of drug development. For example, the Tissue Triage system improves process efficiency in preclinical toxicology by optimising reporting of 80% of "normal" tissue images, thereby reducing the study reporting cycle from a few months to a few weeks. This gives pathologists more time to focus on "abnormal" findings, which also translates into faster reviews.
The company’s Spermatogenic Staging solution optimises staging assessment of testis images in rodents from a few days/ weeks (depending on the expertise of the pathologist), to 15 minutes per image, when assisted by AI. Finally the Predictive Toxicity models predict long-term toxicity of a compound based on short-term studies, reducing, for example, a six-month study cycle to a month or less, and translating into savings in terms of time and resources, as well as number of animals sacrificed. The AI-based solutions once used in drug research reduce the need for animal sacrifice, and provide a boost to humane animal research in line with the 3R principles of animal testing (Reduce, Refine and Replace).
Increasingly, organisations are investing in scanning machines because of the benefits of digital pathology. They recognise the benefits of storage of samples without any deterioration, the ability to collaborate remotely, tele-pathology, leading AI-based solutions and overall work process and diagnostic accuracy improvements.
There are a number of companies in the diagnostics space that do similar work — offering digital analysis of images for diagnosis, mostly products based on telepathology. “A digital collaboration module that can be applied for ‘telepathology’ to provide remote diagnosis or expert consultations is one of the solutions that we provide.” says Kondragunta. Diagnosis is still provided through manual examination and analysis by pathologists. In contrast, AIRA Matrix’s major focus is on providing deep learning-based solutions that actually automate analysis and act as decision-support tools for pathologists, and help improve accuracy, reproducibility and turnaround time of reporting. These solutions perform image analysis and provide diagnostic and prognostic information to support pathologists’ reports. The company provides vendor-neutral software products and solutions for image analysis and management. It currently has tie-ups with 10 companies.
Three pathologists and more than a dozen consultants, microbiologists and 40 people are engaged in AI or analytics-based work at AIRA Matrix.
