As the capabilities have been proven, AI is establishing a firm foothold in the key areas of R&D, drug development, clinical trials, and, to some extent, patient-facing products.
Artificial intelligence is becoming the darling of the biopharma industry after years of testing and discussion. As the capabilities have been proven, AI is establishing a firm foothold in the key areas of R&D, drug development, clinical trials and, to some extent, patient-facing products.
Announcements of new AI deals, partnerships and product launches in biopharma are coming at a rapid pace. A new report from Research and Markets, AI In Pharma Global Market Opportunities and Strategies to 2030: COVID-19 Growth and Change, predicts this segment will grow from almost $699.3 million in 2020 to more than $2.895 billion in 2025. That would result in an annual growth rate of nearly 33%. Afterward, the annual growth rate is predicted to settle to just under 26% for a $9+ billion market by 2030.
This growth is based upon AI’s ability to analyze massive quantities of data and identify important, but often overlooked elements, and to predict outcomes based on historic analyses. Here are a few that BioSpace finds interesting.
Clinical Trials
In clinical trials, drug developers are using AI to extract information from real-world evidence and to make predictions, Deloitte noted in a 2021 Life Sciences Connect podcast. The benefits include shorter trials, lowered costs and better outcomes and productivity.
Perceiv Research Inc. just announced the development of Foresight AD™, an AI-based tool to predict the progression of Alzheimer’s disease patients. In making that announcement, senior scientist Angela Tam, Ph.D., noted that such prognostic models “can significantly enhance clinical trials’ sample quality, reduce trial costs, and decrease the likelihood of endpoint failure.”
Genomenon, an AI-driven genomics company, just announced a proof-of-concept agreement with Deep 6 AI, an AI-based clinical trial acceleration software company. This collaboration supports precision matching for a central nervous system (CNS) study.
In September, Genomenon signed an agreement with Alexion, AstraZeneca Rare Disease to create genetic data sets for a group of rare diseases that includes Wilson disease, complement-mediated thrombotic microangiopathy (CM-TMA), lysosomal acid lipase deficiency (LAL-D) and hypophosphatasia (HPP). The goal is to help patients with rare diseases be diagnosed accurately and, ideally, quickly.
Drug Development
In drug development, Healx, a British biopharma company, is beginning a Phase IIa clinical trial to manage the symptoms of Fragile X syndrome. The compound was discovered using AI.
Earlier this fall, C2i Genomics announced an agreement with NuProbe Global to scale its AI-powered cancer intelligence program throughout China and the U.S. to enable more accurate cancer screening and monitoring and to gain greater insights into the dynamics of the disease.
Likewise, Isomorphic Laboratories, a spinout from Alphabet (the parent of Google) just announced its intention to use AI for drug discovery. The work is based on the AI work of sister company DeepMind in predicting the structure of proteins directly from its amino acid sequence with atomic-level accuracy.
R&D
Deepcell uses AI for cell classification and isolation in basic and translational research. It recently announced a collaboration with the University of Zurich to identify and sort rare melanoma cells, analyze those cells at the molecular level and profile melanoma tissues. The project is expected to enhance understanding of the tumor microenvironment.
“Melanoma cells are difficult to isolate with conventional sorting methods because they lack reliable cell surface markers,” Mitch Levesque, associate professor, University of Zurich, said in a statement. “By isolating and sorting cells using morphology, we may deepen our understanding of the biology of melanoma progression and, in particular, of cell phenotypes and molecular features of cancerous cells.”
Earlier, in September, Deepcell announced an agreement with Stanford University to contribute data to Tabula Sapiens, a ground-breaking program that plans to develop a “benchmark human cell atlas.” It will involve two million cells from 25 organs of eight people. The goal is to create a detailed look at cell types and their distribution and variation across tissues and within the endothelial, epithelial, stromal and immune compartments.
Consumer-Facing Products
For consumer products, Mira uses AI to analyze and make personalized suggestions to women using their hormone tracking kits, to help them become pregnant. Knowing exactly when ovulation occurs can make the difference between getting pregnant and not. Determining ovulation usually is done by at-home kits that measure luteinizing hormone (LH) but, afterward, there has been a lack of clarity about the predicted fertility window. Mira provides that clarity with a new kit to monitor levels of pregnanediol glucuronide (PdG) in the urine by measuring progesterone, a urine metabolite.
The kit, the Progesterone (PdG) Confirm Wand, confirms ovulation and sustained luteinization between ovulation and pregnancy. It recently gained FDA registration as a Class I (minimal to low-risk) over-the-counter tool for women trying to become pregnant.
What makes Mira’s kit particularly novel is its use of AI. “The Mira Analyzer uses machine learning to track changes in a woman’s hormone levels over time, analyze the data and give suggestions about an individual’s reproductive health,” Mira CEO Sylvia Kang told BioSpace. “It’s getting smarter, and also broader, so it’s not just tracking vital signs.”
“The Analyzer is part of the kit,” Kang said. Data goes from the wand to the analyzer, where it’s read. Then it goes to the cloud for AI-based analysis. Hormone data, testing data, the trends curve and the menstrual cycle data is then sent to the user’s cell phone. “Based on the data and the trends in the woman’s own cycle, the AI suggests steps women should take at each point of their cycle to become pregnant, and specific relevant reading material,” Kang added.
The AI element is important, she said, because “although tracking gives women their data, everyone is very different. We’re not made to standards, and that confuses a lot of users. They need to understand what this data means to me. That’s where AI comes into play – it interprets the data and interpersonal trends for individual users.”
