We’re on the cusp of realizing true augmented intelligence. In data-heavy industries with a lot at stake, AI could reduce cost, operationalize the process, assist in scale, and reduce error. Nowhere is this more apparent than the process of drug discovery. AI in Biotech and pharmaceuticals can help because drug discovery is slow. It’s expensive. It’s laborious and requires years (if not decades of research) ending in clinical trials in which only 10% of products make it. You’d almost be better off playing the lottery.
[Related Article: ODSC East 2019: Major Applications of AI in Healthcare]
We love stories of discovering treatments “by chance,” but in reality, those were the results of years of misguided efforts for a different outcome. Sure, now we can regrow hair, but where does that leave us with hypertension?
Luckily, AI is enabling faster drug discovery through better, more accurate data processing and prediction capability, but we’d be remiss if we focused only on this narrow field of Biotech. In reality, AI can be applied to nearly every part of pharmaceuticals, making it a far more interesting (and valuable) commodity than we thought. Here are some things you must be paying attention to in the Ai-Biotech mashup
We can’t get away without talking about drug discovery, but in reality, this process is hype-level exciting. Drug discovery takes decades, wasting $2.6 billion on clinical trials with pretty dismal results. We can spin it all we want (serendipity in drug discovery?), but it’s slow and often demoralizing, turning scientists into financial experts and Pharmaceutical companies into villains.
AI is so much better at using things like existing patient data and computer vision to simulate chemical interactions and predict behavior based on genomics. Adam did it with yeast genes. Adam’s more advanced counterpart Eve discovered that an ingredient in toothpaste could be leveraged to fight drug-resistant malaria parasites.
ODSC even hosted a talk about AI advancing Alzheimer’s research. It’s speeding up discovery time, reducing overall cost, and using predictive models rather than pure trial and error – way better than serendipity.
For a real-world example of how AI is changing this field, hear from Dr. John Reynders, VP, Data Sciences, Genomics, and Bioinformatics at Alexion Pharmaceuticals, Inc. In his talk at ODSC East 2020, “Passing the Turing Test in Rare Disease Diagnosis,” he will explore how Alexion Pharmaceuticals is using AI-based approaches in the field of rare diseases.
Drug discovery is one thing, but biotech companies are hoping to utilize AI to provide more personalized medicine in the next decade. Personalized medicine can help reduce side effects and ineffective results, creating better patient outcomes and fewer recalls or lawsuits.
Johnson & Johnson and Pfizer are already using IBM Watson to analyze a treasure trove of patient data stored in the cloud to recommend better treatment options with a natural language system. Companies like Chilimark research strongly suggest that truly personalized medicine isn’t even possible without machine and deep learning.
Deep generative models create realistic samples from training data, getting around medical data silos, for now. As a field with high privacy walls and deeply ingrained segmentation, this could prove vital to filling in information gaps.
And as the cost of genome sequencing falls, we could find ourselves analyzing the data from a single patient to provide micro-targeted treatments without placing an undue burden on physicians. Augmented intelligence could usher in a new age of treatments, democratized for all.
Realistically, this is still a long way off. Genomics comprises some of our most personalized data, and we could see an uptick in regulations making it harder to get access. The cost to run AI models like this are also still prohibitively expensive for all but the most advanced and well-funded institutions, keeping it squarely in the realm of theory—for now.
To learn about genomics in-person, check out the talk from John Mercer, Head of Data Science at Foundation Medicine at ODSC East 2020, “Delivering on the Promise of AI in Precision Medicine Oncology,” as he plans to discuss the promise and power of genomic data to push the boundaries of harnessing AI to transform cancer care.
Biotech companies are hoping to capitalize on AI’s data processing to spot potential conditions sooner and with greater accuracy. Early studies showed AI’s accuracy in detecting a disease state to be one percentage point higher than medical professionals and two percentage points higher in accuracy for all-clear declarations.
This trend has continued with one caveat. In most studies, medical professionals aren’t given relevant case study information or a medical relationship that’s often at the heart of accurate diagnoses. In real-life use cases, medical professionals are still pinpointing accurate diagnoses.
Where AI seems highly promising is on relieving the burden overworked medical professionals have, especially in areas where there’s less access to medical care. As doctors are allowed less time with patients and patients won’t (or can’t) receive regular care, the data processing capability of AI stands out.
Companies are also exploring using patient-centered analytics as a way to round out and aid what doctors know from their charts. Strava, for example, uses advanced analytics to provide athletic data for athletes. These could also be sources of vital information in the future for diagnostics and medicine. Hear how Strava does it at ODSC East with Cathy Tanimura, Senior Director of Data Analytics in her talk, “The Power of Visualization: Best Practices for Effective Visualizations.”
The research hasn’t always panned out, but with careful consideration and better use case examples (the partnership between IBM Watson and Memorial Sloan Kettering, for example), we could see AI augmenting diagnoses and reducing errors and omissions.
Non-Medical Related AI
AI is also affecting largely unseen aspects of BioTech and Pharmaceuticals. Things like marketing, lab assistants, and administrative work take up considerable resources. Marketing is a direct contributor to the success of new drugs, at least in the United States, and AI could provide better insights into those campaigns.
Like other AI-driven marketing, big data is giving companies better predictive capabilities and the chance to process unstructured data to create pictures of customer journeys. Since marketing efforts are spread across a wide variety of media, from print to commercials to social media, AI could sift through the data for actionable insights.
AI can also assist in administrative tasks like lab work. For example, Desktop Genetics created a platform designed to build CRISPR libraries through AI. It allows even those with no experience using CRISPR to expedite the process without sacrificing the integrity of the guides.
Other startups are focused on easing the data bottleneck, making sure that rich analysis exists from the volume of healthcare data available. At the last estimate, lost data costs the industry around $100 billion each year. AI could reduce that number.
A Lot of Hype and a Lot of Promise
[Related Article: 15 Open Datasets for Healthcare]
Right now, much of AI’s use cases in both BioTech and Pharmaceuticals are hype, but the tide is turning. As AI grows more intelligent, we could see streamlined processes for drug discovery, the diagnostic tools of our dreams, and even a proliferation of AI to reduce the administrative burden of research. 2020 is only going to get more interesting.
Ready to learn more about the use of AI in medicine and biopharm? Be sure to check out all of the biopharm and healthcare-related talks at ODSC East this April 13-17. More highlighted talks include:
- “From Data Strategy to Deep Learning: Enabling AI Solutions for Life Sciences & Pharmaceuticals” with Michael Segala, PhD, CEO of SFL Scientific. The goal of this session is to demonstrate and learn about building end-to-end AI pipelines for biotechnology and healthcare applications.
- “Strategies for Building AI-ready Data Sources and (Semi)autonomous Reasoning Agents Operating on Top of Them” with Marcin Grotthuss, PhD, Computational Biologist at The Broad Institute of MIT & Harvard. This talk will discuss the strategies of how to integrate and provide high-value AI-ready data sources as well as how to develop (semi) autonomous reasoning agents that would advance reasoning through innovative uses of these knowledge sources.
- “Convergence and Critical Mass: The Fusion Moment for Biopharma” with Brian Martin, Head of AI at AbbVie. This talk will lay the groundwork explaining why for the sake of biopharmaceutical research and for human life we must transform the way the industry sees and uses data.