Deep aging clocks – the emergence of Al-based biomarkers of aging and Longevity and their implications in the development of therapeutic restorative interventions that can increase lifespan and healthspan.
Aging does not only entail the deterioration in physical appearance; the organism’s ability to regulate physiological functions decline as well. Understanding aging for an is not a ‘2 dimensional’ question as it’s multifactorial and more akin to a Rubik’s cube. As the recent report from Aging Analytics Agency ‘Biomarkers of Longevity’ noted:
“As AI for R&D in drug discovery becomes more sophisticated, drugs will become more customised to specific diseases and even specific patients. Drug development companies will transition from the current form of “blockbuster drugs” (standard drug formulations applicable to many millions of patients) to P4 medicine, tailoring drugs to specific patient cases based on age, gender, ethnicity, state of health and genetics.”
Aging is associated with several diseases, including diabetes, cancer, heart diseases, and neurodegenerative disorders. Consequently, interventions promoting Longevity are also expected to prevent such age-related conditions. The discovery of robust aging biomarkers is of high importance, without which the development and assessment of Longevity-promoting therapies would not be possible (and investible).
Specific epigenetics markers (also known as Horvath’s Clock) were among the first biomarkers of aging and paved the way for the development of novel and more robust ones . Due to the recent progress in artificial intelligence (AI) and machine learning (also known as deep learning), computer systems can perform complex tasks in a very efficient and reliable way. For example, deep-learning systems can be trained to measure simple features that are changing in time, and then apply these principles to measure changes and make predictions in more complex biological processes, including aging .
The deep age predictors (also known as deep aging clocks) have emerged as a powerful tool that can foster advances in aging research and the development of Longevity-promoting interventions. They can contribute to the identification of novel therapeutic targets and the assessment of the efficacy of aging-targeting therapeutic interventions. Deep aging clocks can also be used in biological data quality control, genomic data analysis, data economics, and prediction of the presence or onset of age-related diseases.
Referring to his paper titled “Deep Aging Clocks: The Emergence of AI-Based Biomarkers of Aging and Longevity” in Cell Trends in Pharmacological Sciences, which he co-authored with Polina Mamoshina, Senior Scientist at Insilico Medicine, Alex Zhavoronkov, the Founder of Insilico Medicine, commented:
“Humans are very good at guessing each other’s age using images, videos, voice, and even smell. Deep neural networks can do it better and we can now interpret what factors are most important. Very often when someone looks older than their chronological age, they are sick. A trained doctor can guess the health status of a patient just by looking at him or her. At Insilico we developed a broad range of deep biomarkers of aging that can be used by the pharmaceutical and insurance companies, as well as by the longevity biotechnology community.” 
The aging clocks can help us obtain a deep understanding of the changes that happen during aging and the underlying mechanisms. Only then we will be able to develop preventive and therapeutic restorative interventions that can increase lifespan and healthspan by reversing the biological clocks back to the young productive, healthy state. The leaders in the field expect this trend to continue with the inevitable commercialisation in the insurance and consumer wellness industries.
Commenting on biomarkers, Lara Puhlmann of at The Max Planck Institute for Human Cognitive and Brain Sciences said: “I would like to look at further neurological markers of aging. For example, there are several promising computational approaches for predicting a person’s ‘brain age‘, and deviation from the ‘normal’ or ‘ideal’ brain age. These ‘brain predicted ages’ appear to be clinically meaningful and, as a concept, are quite accessible, which makes interpretation easier. Alternatively, there are also some interesting findings with epigenetic clocks. Ideally, one would combine these multiple approaches and use them to cross-validate each other, or integrate them.”