Recent advances in deep learning enabled the development of AI systems that outperform humans in many tasks and have started to empower scientists and physicians with new tools. In this Comment, we discuss how recent applications of AI to aging research are leading to the emergence of the field of longevity medicine.
Aging is a universal feature shared by all living beings. While the rate of aging may vary among individuals and species, the time elapsed since birth is a strong predictor of health status and mortality. Targeting aging may extend the average life expectancy more substantially than prevention or treatment of individual diseases1. However, within the established drug discovery and development framework, pharmaceutical companies are still searching for compounds and interventions for the treatment of individual chronic diseases such as cancer and cardiovascular or pulmonary diseases. Current biomedical research aims to identify the underlying mechanisms and molecular targets specific to a disease in order to modify the disease, treat its symptoms or cure it. Once a drug is approved, the medical community is obliged to follow a specific and defined protocol that is mostly targeted towards treatment of a specific disease with a specific drug or a combination. Rarely, however, do clinicians prescribe an off-label drug, even when there is evidence that it may be effective in the treatment of another disease out of concerns about possible side effects and to avoid possible malpractice claims. While evidence-based medicine has been overwhelmingly effective at reducing mortality in the past half century, it has also increased the economic burden of disease in developed countries due to the resulting extension in lifespan without a concomitant increase in healthspan. Since aging plays a key role in the onset and progression of many human diseases and affects all organs of the body, many chronic conditions manifest themselves simultaneously as comorbidities later in life. Combined with global trends in population aging, this apparent victory of modern medicine to extend our lives has led to an explosion in healthcare costs and has failed to concomitantly improve wellness and quality of life for older adults
The advent of artificial intelligence in biomedical research and medicine
Preventive, geroscience-based measures to treat aging at the organismal level would provide a more substantial benefit than reactive therapeutic approaches targeted at a single disease or organ, since those do not contribute to a significant improvement in healthspan. It is often ignored that disease-focused therapeutic approaches do not remarkably extend lifespans. It has been estimated that the complete elimination of a single fatal disease such as cancer in the USA would merely lead to a 2.3-year population increase in life expectancy at birth and a 1.2-year increase at age 65, since the majority of overall mortality is due to age-related diseases and the abovementioned multimorbidy1. Old biological age is not recommended to be reported as an official cause of death, but it remains the main reason why older adults die worldwide. Therefore, combining the prevention and elimination of chronic diseases by adding a geroscience approach to routine clinical care would yield the best outcomes in that it would promote both a long and healthy lifespan.
However, understanding the aging process requires longitudinal monitoring of millions of parameters in many different types of data sets that change very slowly during the human life course, and distinctly in genetically and socioculturally diverse populations3. While humans can be trained to accurately predict age and explain features leading to their predictions — using facial images, for instance — and even to propose corrective actions, no human doctor can do this on multiple different biological data types, such as blood tests or gene expression data4. Fortunately, the task of finding complex patterns in large volumes of longitudinal data is where modern artificial intelligence (AI) demonstrates a unique, often spectacular performance. Since 2014, AI systems have outperformed human experts in multiple areas, including image recognition, knowledge quizzes, video games, language translation and many other tasks. Most of these achievements have been made possible by various advances in deep neural networks trained on large data sets using high-performance computing. Importantly, deep generative reinforcement learning has been successfully used in a broad range of biomedical applications ranging from drug discovery to prediction of clinical trial outcomes and personalized medicine.
AI-powered tools enable longevity medicine
Deep learning (DL) was a breakthrough for AI research, allowing for the training of deep neural networks (DNNs) on massive longitudinal data sets, which were previously almost impossibly difficult to comprehensively mine and interpret in the longevity arena. There is no consensus on how to define human ‘biological age’, but the term usually refers to a measure that is more predictive of mortality, diseases or frailty than chronological age, and one that changes in response to geroprotective interventions and can track some of the biological hallmarks of aging, such as the promotion of cellular senescence by smoking, for instance. DL was instrumental in establishing Deep Aging Clocks (DAC) that generate estimates of an individual’s biological age state based on data extracted from routine blood analyses. Using AI-powered tools such as DAC, clinicians should be able to more precisely assess and monitor individual health risks and tailor appropriate interventions or changes in lifestyle for a specific person. We argue that DAC should become an essential part of the physician’s tool kit, enabling AI-supported recommendations to promote long and healthy lives. Other DL-based solutions that outperform humans, such as radiological image analysis algorithms for early cancer or aneurysm detection as well as dermatological testing, can provide complementary benefits. In this context, we define longevity medicine as a branch of precision medicine that is specifically focused on promoting healthspan and lifespan, and is powered by AI technology. AI-powered longevity medicine will facilitate the discovery of drug targets for specific individuals, the identification of tailored geroprotective interventions and aging and longevity biomarkers to enhance the study of aging and disease trajectories, and the identification of interventions that may help slow down or even reverse aging-associated biological, physiological or psychological processes. As recent advances in longevity biotechnology and AI are starting to percolate through clinical research and clinical practice, physicians will increasingly need to navigate through various AI technologies and applications, including those that may be relevant to the nascent field of longevity medicine.
Opportunities for longevity medicine in clinical care and the longevity industry
In order for longevity medicine to be formally seen as a branch of medicine, it needs to be practiced by physicians....
Original article https://www.nature.com/articles/s43587-020-00020-4