Artificial Intelligence (AI) has positively impacted many industries including healthcare, transportation, e-commerce, and finance. According to a recent report, the global artificial intelligence software market is forecast to grow rapidly in the coming years, reaching around 126 billion U.S. dollars by 2025. Industry leaders have come to recognize the benefits of AI with its ability to significantly reduce time needed to perform a task by automating workflows, provide insights to make smarter decisions, and reduce the margin of error.
AI enables us to recognize patterns and insights in vast amounts of structured and unstructured data, which leads to faster, better, and more unbiased decision-making across all sectors. Industries based on scientific and technical disciplines are faced with an especially large volume of disparate and complex information given the amount of data that is generated by R&D, and therefore are able to benefit greatly from the application of AI. Over the last decade, advances in the life sciences industry in particular have been fueled by an exponential increase in large data sets amenable to machine learning (ML), vast improvements in computing power, and more comprehensive software libraries that can execute the relevant tasks. The results of these technological leaps are seen all across the life sciences value chain, from early drug discovery through clinical trials, supply chain and manufacturing, and commercialization.
The Growth of Data
Professionals in the life sciences industry are overwhelmed with data, making it difficult if not impossible, to navigate the tsunami of information available to them. Current estimates of the number of scientific publications and abstracts in the biological and medical sciences on PubMed alone is 33 million, and the volume of scientific research has a predicted annual growth rate of 4.1%, doubling every 17.3 years. In parallel, the industry has seen a substantial increase in “omics” data sets (e.g., genomics, transcriptomics, proteomics, metabolomics) as well as patient data from clinical trials, electronic health/medical records (EHRs/EMRs), and wearables/monitors.
The net result of this data explosion is that it is fertile ground for AI technologies that leverage ML, including deep learning (DL) neural network approaches. This opportunity is not without challenges, however, many of which cut across sectors. A key issue is that the data takes the form of both structured and unstructured information, which needs to be carefully curated and normalized to make the right connections. For instance, a clinician and an investor may use different terms to describe the same concept, a difference that is known as a “semantic gap”. These differences need to be resolved computationally, and advances in natural language processing (NLP) have been instrumental in addressing this issue. Another challenge relates to explainability of the insights that an AI produces: the second wave of AI applications using approaches such as neural networks suffered from “black box” syndrome, in which the user was not able to understand intuitively how and why a particular outlier or finding was important and relevant. Fortunately, explainability is a focus area for AI application development and we will continue to see major advances as systems autonomously learn and deploy logical reasoning to recognize patterns. Biology and medicine are filled with interconnected mechanisms, diseases, and processes, as well as significant gaps in our understanding of their relationships: AI’s chief goal is to turn all of this data into knowledge and meaningful insights.
Artificial Intelligence in the Life Sciences Industry
The life sciences and healthcare industry is comprised of four main sub-sectors, namely therapeutics, diagnostics, medical devices, and healthcare IT (HCIT). Recent data from KPMG shows that 82% of executives in all of these sub-sectors want to see their companies adopt AI technologies because the benefits of artificial intelligence have already been observed in drug development, diagnostics, clinical trials, supply chain, and commercial and regulatory processes. The McKinsey Global Institute has estimated that AI in the pharmaceutical industry could generate nearly $100B annually across the US healthcare system.
Propelling Research of New Drugs and Products
AI technology is creating a positive disruption in the development of new medical products and drugs. Learning algorithms provide insight and analysis on structured and unstructured data to identify new symptoms of diseases and generate predictions on which drugs are most likely to succeed in treating which patient populations.
Applying AIin research and early drug discovery is an area of intense activity, and there are many examples of its use. Machine learning tools are used to analyze vast omics datasets to gain knowledge of the underlying disease pathways as well as the biomarker profiles of patients who have a disease or may benefit most from a specific treatment. For example, deep learning capabilities have helped in cancer research by predicting neoantigens or tumor specific antigens within cancer cells, which in turn indicates to drug developers which proteins to target. Biopharma companies also use AI-enabled drug discovery platforms to identify new drugs, and refine or repurpose previous drugs in the creation of vaccines and other medicines. AlphaFold is DeepMind’s algorithm designed to predict protein folding, a fundamental mechanism in determining a protein’s conformation (i.e., shape). This knowledge is a key enabler of structure-based drug design: by knowing what the protein’s conformation is, it is easier to design drugs that can recognize and bind to that protein.
Advancing Diagnostics
The technology helps when it comes to crucial decision making, management, and automation by working to considerably reduce human error and provide an accurate diagnosis. Dating back to 2014, one use of artificial intelligence is to act as a second pair of eyes in diagnostic imaging. Using algorithms to quickly and sensitively analyze x-ray, MRI (see Ezra, a company that detects cancer early, and Subtle Medical, using deep learning to enhance image quality and allow faster workflows), and CT images, AI can guide doctors in making a medical diagnosis and selecting an appropriate treatment, ultimately saving precious time and delivering life-changing medical care. In addition, deep learning capabilities can discover previously unknown, meaningful patterns in the data. One notable example of diagnostic and clinical support is IBM’s foray into AI for healthcare known as Watson for Oncology: the platform offers a number of options, from diagnostic imaging solutions to clinical study design and management.
Healthcare Technology
AI has also been an important enabler when it comes to sifting through electronic medical records. EMRs are notoriously heterogeneous since they contain both unstructured information such as healthcare provider notes and structured information such as lab results, prescriptions, and treatment records. MedKnowts is a smarter EHR platform developed by researchers at MIT and Beth Israel Deaconess Medical Center that uses machine learning to speed up physician workflows by auto-populating and surfacing relevant clinical notes based on what the user is looking for. GE Healthcare recently announced Edison Digital Health platform to effectively deploy clinical workflows, analytics, and AI tools to improve the delivery of care in hospital systems.
The Future of AI in Life Sciences and Healthcare
It is clear that artificial intelligence will continue to benefit the life sciences industry given its successes to date and the enormous opportunity to have a transformative impact on our understanding of diseases and the treatments, products, and processes. There are still large inefficiencies and high failure rates in the life sciences and healthcare industry that scientists, clinicians, executives and investors are keen to resolve. By surfacing key insights, AI can assist and extend our ability to identify where to direct our intellectual, physical, and financial capital to maximize the effect and save lives.
Michelle Kim-Danely is a female scientist turned biotech executive. Michelle possesses an extensive background of knowledge in all things life science with experience as a researcher at Roche before starting at ALPHA10X. Michelle has a passion for seeking ways to de-risk opportunities to lead to impactful outcomes.
