Artificial Intelligence for Clinical Medicine
Jeff Morris
Advances in artificial intelligence, including deep neural networks, have direct application in finding aging biomarkers and discovering genetic predisposition to disease. These latest advances in AI and their application for development of aging biomarkers will be the subject of a lecture by Ivan Ozerov, Ph.D., at the upcoming 24th Clinical Applications for Age Management Medicine Conference in Tucson, Arizona. Methods based on deep architectures have outperformed classical approaches, said Dr. Ozerov, not only in image analysis, but also in solving a wide range of complex genomics, transcriptomics and proteomics problems. The session, "Artificial Intelligence for Aging Biomarkers & Age Management Research," will focus on machine learning approaches and deep learning techniques for the development of aging biomarkers, as well as their use for customer-oriented age-management systems that utilize multiple AI-driven approaches, stacked into an ensemble and trained on multiple medical and biological data.
|
| |
Dr. Ozerov is Director, Drug Discovery, Aging and Age-related Diseases, at Insilico Medicine, a Baltimore-based firm engaged in next-generation artificial intelligence and blockchain technologies for drug discovery, biomarker development, and aging research. Dr. Ozerov graduated from Lomonosov Moscow State University in 2010 and defended his Ph.D. in radiobiology in 2015. His Ph.D. thesis was related to kinetic modeling of the cellular pathways affected by exposure to ionizing radiation and cellular senescence. After joining the international team of Insilico in 2015 through winning the Bioinformatics Hackathon in Russia, he developed a novel state-of-the-art tool for omics data analysis on the level of cellular signalling pathways (iPANDA). The team Dr. Ozerov leads has successfully developed and launched Young.AI, an AI-empowered platform integrating multiple biomarkers of human age in order to manage personal health, track changes over time and optimize people's lifestyles. He will discuss the techniques used by platforms such as Young.AI and their clinical use for age management medicine.
Alex Zhavoronkov, Ph.D., who as the founder and CEO of Insilico recruited Dr. Ozerov, said, "Assessing the biological age of the patient using multiple data types may significantly contribute to personalization in many areas of medicine. People, their organs and systems age at different rates and adjusting the therapy to the biological age of the patient may help improve outcomes in clinical trials and in the real world. The deep neural networks trained to predict the biological age of the patient may be used to discover novel targets and pathways in aging and age-related diseases." In addition, said Dr. Zhavoronkov, "Aging is a highly multi-modal process, which deconvolutes into many age-related pathologies and leads to the loss of function. The Young.AI system is intended to take full advantage of the advances in deep learning to track the aging processes at every level of organization, evaluate the importance of each feature within every data type, and to look at the big picture and identify the effectiveness of different interventions."
In September of last year, Insilico announced the launch of the Beta 1.0 version of Young.AI. The first version, publicly unveiled at the 4th Aging Research for Drug Discovery Forum and the Artificial Intelligence and Blockchain for Healthcare Forum in Basel, Switzerland, incorporates predictors of age as well as the ability to track drug and supplement intake. Future updates will include multiple other data types ranging from medical imaging and brain activity readings to social circle and behavior. The system allows users to register under a pseudonym and remain anonymous, while entering the available data types and tracking the age predicted by the deep neural networks trained on tens of thousands and sometimes millions of samples.
Blockchain technology enables the creation of a distributed and secure ledger of personal data, with patients in control, owning their data, monitoring access privileges and understanding who looked at the data. “Most importantly,” said Dr. Ozerov, “blockchain technology allows for the creation of a data-driven marketplace, where patients can earn tangible rewards for making their data available to the application development community, pharmaceutical and consumer companies, and research institutions, and generating new data through regular and comprehensive tests and checkups. Presently, only a few patients worldwide have the comprehensive data sets containing their clinical history combined with the genetic, blood biochemistry and cell count profiles, lifestyle data, drug and supplement use and other data types, because they do not see the value in this data and do not get tested regularly. On the other hand, pharmaceutical and consumer companies alike are willing to pay substantial amounts for the large numbers of personal data records required to train their AI. These funds can be used to subsidize regular testing by the patients, uncover new uses for the various data types and develop sophisticated diagnostic and treatment tools.”
Vadim Gladyshev, Professor of Medicine at Brigham and Women's Hospital, Harvard Medical School, was quoted as saying about Young.AI, "The use of the new tool to track human biological age may enable discovery of drugs and other interventions that target the fundamental process of aging, thereby delaying the onset of all chronic diseases at once, instead of targeting one disease at a time."
Peter H. Diamandis, M.D., founder and chairman of the X Prize Foundation and cofounder and executive chairman of Singularity University, said of the technology, "Genomics, cellular therapeutics, CRISPR/Cas9 technologies are all giving us tools to extend the healthy human lifespan. Today many billions of dollars of private investment are going into the longevity start-up space, and people are starting to think about aging as plastic and moldable. Having technologies like Young.AI to objectively measure the biomarkers of aging will be an important part in humanity's tool kit. You can't fix what you can't properly measure."
A colleague of Dr. Ozerov, Senior Deep Learning Scientist Polina Mamoshina, said, "Most of the research in biomedicine and in the anti-aging field in general is done in animals. Even some of our validation experiments rely on more primitive organisms. However, if we are to develop actionable interventions that can prevent the onset of Alzheimer's, Parkinson's, fibrosis, CVD and metabolic diseases and extend the youthful state of the human body, we need a very comprehensive and sensitive system for biomarkers, and this can only be developed by tracking a very large number of people over time. Presently, we do not understand the value of each data type to aging research, and Young.AI may help us pick up the low-hanging fruit and create a platform for further research." She added, "If we are to develop actionable biomarkers of aging, we need a comprehensive and robust approach. Such an approach can only be developed using a large number of samples from multiple populations. We are working on multiple biomarkers using deep learning and incorporating blood biochemistry, transcriptomics, and even imaging data to be able to track the effectiveness of the various interventions we are developing".
A paper published by a group of Dr. Ozerov’s colleagues in January’s Journal of Gerontology, "Population-specific biomarkers of human aging: a big data study using South Korean, Canadian and Eastern-European patient populations," demonstrated the application of deep neural networks to assessing the biological age of the patients.
In the paper, the authors present a novel deep-learning based hematological human aging clock, a biomarker that predicts the biological age of individual patients. The study uses a large dataset of fully anonymized Canadian, South Korean and Eastern European blood test records to train an aging clock. The developed model predicts age better than models tailored to the specific populations highlighting the differences of subregion-specific patterns of aging. In addition, the developed clocks were shown to be a better predictor of all-cause mortality than chronological age. The paper includes co-authors from Gachon University Gil Medical Center, University of Copenhagen, University of Alberta, and the Biogerontology Research Foundation.
The paper states that the most accurate methods of calculating biological age is a subject of ongoing debate, and recent studies suggest that a suite of biomarkers, rather than any individual biomarker, constitute the most effective means of assessing the health status of a patient. In deciding to focus on emerging machine learning techniques, such as deep neural networks (DNNs), in the construction of their aging clocks, the study authors noted that DNNs are perceived as game-changing methods in data analysis due to their capacity to capture hidden underlying features and learn complex representations of highly multidimensional data.
One of the paper’s co-authors, Franco Cortese, Deputy Director of the Biogerontology Research Foundation, said, "Development of effective biomarkers of age is one of the most pressing goals in geroscience today, as it lays the foundation for efficient preclinical and clinical evaluation of potential healthspan-extending interventions. Humans live a long time, and testing the effect of gerontological interventions in humans using lifespan gains as the main criterion for success would be wildly impractical, necessitating long and costly longitudinal studies. By developing accurate biomarkers of aging, the efficacy of potential healthspan-extending interventions could instead be tested according to changes in study participants' biomarkers of age. While significant attention is paid to the development of highly accurate biomarkers of aging, less attention is paid to developing actionable biomarkers of aging that can be tested inexpensively using the tools at hand to the majority of researchers and clinicians. We developed the deep-learning based, blood biochemistry aging clock presented in this paper in the hopes of making progress toward the goal of more actionable biomarkers of aging."
"This work demonstrates the synergy between artificial intelligence and aging research,” said Lee Uhn, Ph.D., Chief of Artificial Intelligence at the Gachon University Gil Medical Center. “Every living being has age as a feature and it is possible to engage in multi-national collaborations using the very simple data types to assess the population specificity of age predictors. Our group is using advanced AI for multiple clinical applications and has a working collaboration with IBM Watson."
Dr. Ozerov says his and his colleagues’ work may help improve clinical trial enrollment practices, assess the population specificity of a variety of biomarkers and pave the way for the development of more complex multi-modal biomarkers of aging and disease.
For the latest information on the 24th Clinical Applications for Age Management Medicine Conference in Orlando, visit www.agemed.org.
Return to March 2018 e-Journal 