Longevity Conferences 2023
Curated list of Longevity Conferences, where you can explore the latest research and developments in the field of aging and longevity.
Modern machine learning-based AI shows great promises in the medical field, particularly for tracking healthy aging.
Artificial intelligence (AI) as a concept was formulated already in 1956 by John McCarthy and can be formulated as computer algorithms mimicking features of human intelligence, such as learning and problem-solving. For the last decade, the tremendous growth of computational power and data availability paved the way for the widespread use of AI across all fields, and the medical field is no exception.
The current applications of AI are based on machine learning (ML) techniques – complex statistical models that can extract dependencies from the data without being explicitly programmed, make predictions, and some are able to learn new information. An example of the ground-breaking powers of AI is AlpaFold by DeepMind (1), which solved the fundamental problem of protein folding. ML provides an attractive alternative to other approaches in areas with little prior knowledge about those dependencies or where they are too complex. Current applications and successes of ML largely rely on available data volumes. The growth of data continues, and, as of today, the term big data refers to petabytes (1024 terabytes) and exabytes (1024 petabytes) of information.
The medical field is not excluded from an all-present spread of AI. The so-called digital health revolution resulted in the acquisition of massive clinical data in a diverse population. The volume and variety of data obtained through electronic health records and sensor-equipped devices give a vast field for the application of ML. Electronic health record systems allow to improve evidence-care guidance (2), and high-capacity and real-time data processing enhanced by ML give the possibility to provide personalized decisions (3). Also using AI at the bedside allows enhancing human-guided diagnostics and prediction. The Internet of Health Things (IoHT) concept includes smart health objects, data from which unites into one personal record (4). The application of ML together with IoHT may further improve the development of AI personalized systems.
A large emphasis is placed on improving the quality of care for older adults by implementing AI in general diagnostics and geriatric oncology programs (5). Though still in their early stages, such AI-based platforms hold promise for successful management of the complex conditions in the older population.
A variety of machine learning methods is widely used today, but the method that drives the most attention is the neural network, particularly, deep learning. Artificial neural networks (ANN) were initially created as a model inspired by a human brain, with an input layer that gets information and an output layer that produces a result. Each layer of ANN consists of nodes or neurons, each of which contains some non-linear processing function and possesses some weight. Early ANNs included only input and output layers and did not show a high predictive power. Introducing one or more hidden layers between input and output boosted ANN's predictive power, and modern deep neural networks (DNN) (6) include many hidden layers. Training a DNN revolves around the following:
The power of DNNs is their ability to learn complex abstractions by constructing meaningful features during the learning process (similarly to a human brain). The architecture of DNNs varies and can be incredibly complex, including different types of functions and different interconnections between layers. There are different subtypes of DNN, for example, the Recurrent Neural Networks (RNN), which have been applied in the health area because they are appropriate for the treatment of data such as text, speech, and DNA sequences (7). In this type of network, each state affects the results of the following one — thus presenting a memory in the network.
Despite the vast popularity of DNNs, other ML methods are still widely used and able to provide comparable results. Among the most powerful ones are the methods based on the decision trees – simple classifiers that can be depicted as a set of cascading decisions (if A then B, if not A then C) with multiple levels. Widely popular random forest and xgBoost techniques both use different ensembles of decision trees. In random forest decision trees are independent and model predictions are obtained by “voting”, while in xgBoost each new decision tree is trained to predict the error of the previous one thus minimizing the general error of prediction.
One of the most known applications of AI in diagnostics is IBM Watson Health™, which includes a range of ML models, including DNN ones, and promises high accuracy in the diagnosis of various diseases. IBM Watson learns on unstructured and semistructured data from the clinical literature, health records, and test results identifying the most important pieces of information and then mining a patient's data. The system then forms and tests hypotheses and finally provides a list of individualized recommendations, including a patient's eligibility for specific treatments. IBM Watson Health™ presently offers commercialized applications of the Watson system for genomics, drug discovery, health care management, and oncology (8). However, as with any AI, Watson is not performing equally well in all types of diagnostics. A study on the oncology patients in China showed that for ovarian cancer the concordance was 96% and for lung cancer and breast cancer obtained – slightly above 80%, but for colon cancer and cervical cancer – already lesser than 64%. And, for gastric cancer, the system reached a very low concordance of only 12% (9).
So, despite the hopes of obtaining a universal AI tool for all diseases, there is still a need for specialized AI systems. And such particular use various ML methods found in imaging (10); ophtalmology (11); diagnostic and outcome predictions for multiple types of cancer including breast (12,13) and skin cancer (14); cardiovascular diseases (15); and diabetes (16,17).
AI can also be used to assess aging and general health status not only through the identification and treatment of most widespread age-related diseases but also by detection of health deterioration, physical and cognitive frailty.
DeepSigns (18) is an RNN-based system that allows early detection of deterioration signs that uses the data of ICU (intense care unit) hospitalized patients. It uses the APACHE II index based on vital signs, such as blood pressure and temperature, and results of laboratory exams. Based on the patient's data, DeepSigns can predict future changes in vital signs with an accuracy of around 80%. The test results from DeepSigns identified 50% and 60% of otherwise unidentifiable cases. As 17% of the 50% and 60% tested patients from learning data died, it means that around 9% of deaths could be prevented with the DeepSigns model application to the early intervention in ICU units. The future application might include not only embedding DeepSigns in ICU warning systems but also in-home use for elderly high-risk patients.
Several AI systems showed promising results in addressing frailty - a geriatric condition linked to an elevated risk of rapid declines in health and function among the older population. The degree of frailty is commonly used as a measure of health in aging (19). Frailty depends on a multiplicity of factors, and ML was successfully used to analyze this complex multidimensional data. Gomez-Cabrero et al. (20) employed random forest classification to develop a frailty prediction system. Analysis of the models allowed them to identify both main protective biomarkers – vitamin D3, lutein zeaxanthin, and miR-125b-5p – as well as risk biomarker, cardiac troponin T. Another study by Aponte-Hao et al. (21) analyzed data on more than 5,000 non-frail and frail patients using a range of models including ANN and xgBoost. Authors were able to develop an accurate prediction system, and analysis of models revealed that frail patients were statistically significantly likely to be older, female, and less likely to have no known chronic conditions. Of those with at least one chronic condition, frail patients were more likely to have chronic obstructive pulmonary disease, dementia, depression, and hypertension. The advances in such ML models allow not only to extract patterns of frailty but to develop new indices able to predict short-term mortality as a part of AI integrated systems (22,23).
Aging clocks are the composite systems that can estimate an organism's biological age (age of its cells, tissues, and systems), capture different processes (e.g., cellular senescence) and consequences (e.g., risk of mortality). Such predictors of age based on deep learning are rapidly gaining popularity (24).
Currently developed aging clocks are based on a variety of data; for example, Vineet et al. developed an aging clock based on imaging data, biomarker data (C-reactive protein, glycated hemoglobin, albumin, total cholesterol, and others), physical activity, and anthropometry for more than 20 thousand individuals (25). For each type of data, they developed ML models using various types of neural networks, including DNN with different architectures and were able to build high accuracy predictors and estimate the dependencies between chosen parameters and health/disease status.
Another proposed aging clock, iAge, is based on inflammatory biomarkers (26). From the blood immunome data of more than 1,000 individuals from 8 to 96 years of age, the authors developed a deep learning model based on patterns of systemic age-related inflammation. The iAge is able to track immunosenescence, frailty, cardiovascular aging, and probability of exceptional longevity. Analysis of factors contributing to this model highlighted as one of the most important the chemokine CXCL9 involved in cardiac aging and poor vascular function.
Modern ML-based AI shows great promises in the medical field, particularly for tracking healthy aging, and potentially enables patients' access to the best quality of care. However, the digital revolution is not fast, and full implementation of AI requires both more technical advances and cross-training of clinicians and AI experts (27). The issue of preserving the privacy of patients' data is of great importance, and solutions are being developed (28). The potential inclusion of AI systems into healthcare will allow more precise and personalized treatment for more patients.
References
1. Thornton JM, Laskowski RA, Borkakoti N. AlphaFold heralds a data-driven revolution in biology and medicine. Nat Med. 2021 Oct;27(10):1666–9.
2. Hemingway H, Asselbergs FW, Danesh J, Dobson R, Maniadakis N, Maggioni A, et al. Big data from electronic health records for early and late translational cardiovascular research: challenges and potential. European Heart Journal. 2018 Apr 21;39(16):1481–95.
3. Rajkomar A, Dean J, Kohane I. Machine Learning in Medicine. N Engl J Med. 2019 Apr 4;380(14):1347–58.
4. da Costa CA, Pasluosta CF, Eskofier B, da Silva DB, da Rosa Righi R. Internet of Health Things: Toward intelligent vital signs monitoring in hospital wards. Artificial Intelligence in Medicine. 2018 Jul;89:61–9.
5. Extermann M, Brain E, Canin B, Cherian MN, Cheung K-L, de Glas N, et al. Priorities for the global advancement of care for older adults with cancer: an update of the International Society of Geriatric Oncology Priorities Initiative. The Lancet Oncology. 2021 Jan;22(1):e29–36.
6. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015 May 28;521(7553):436–44.
7. Ravi D, Wong C, Deligianni F, Berthelot M, Andreu-Perez J, Lo B, et al. Deep Learning for Health Informatics. IEEE J Biomed Health Inform. 2017 Jan;21(1):4–21.
8. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, et al. Artificial intelligence in healthcare: past, present and future. Stroke Vasc Neurol. 2017 Dec;2(4):230–43.
9. Zhou N, Zhang C, Lv H, Hao C, Li T, Zhu J, et al. Concordance Study Between IBM Watson for Oncology and Clinical Practice for Patients with Cancer in China. The Oncol. 2019 Jun;24(6):812–9.
10. Barragán-Montero A, Javaid U, Valdés G, Nguyen D, Desbordes P, Macq B, et al. Artificial intelligence and machine learning for medical imaging: A technology review. Physica Medica. 2021 Mar;83:242–56.
11. Hogarty DT, Mackey DA, Hewitt AW. Current state and future prospects of artificial intelligence in ophthalmology: a review: Artificial intelligence in ophthalmology. Clin Experiment Ophthalmol. 2019 Jan;47(1):128–39.
12. Houssein EH, Emam MM, Ali AA, Suganthan PN. Deep and machine learning techniques for medical imaging-based breast cancer: A comprehensive review. Expert Systems with Applications. 2021 Apr;167:114161.
13. Yu K, Tan L, Lin L, Cheng X, Yi Z, Sato T. Deep-Learning-Empowered Breast Cancer Auxiliary Diagnosis for 5GB Remote E-Health. IEEE Wireless Commun. 2021 Jun;28(3):54–61.
14. Khamparia A, Singh PK, Rani P, Samanta D, Khanna A, Bhushan B. An internet of health things‐driven deep learning framework for detection and classification of skin cancer using transfer learning. Trans Emerging Tel Tech [Internet]. 2021 Jul [cited 2021 Dec 23];32(7). Available from: https://onlinelibrary.wiley.com/doi/10.1002/ett.3963
15. Swathy M, Saruladha K. A comparative study of classification and prediction of Cardio-Vascular Diseases (CVD) using Machine Learning and Deep Learning techniques. ICT Express. 2021 Sep;S2405959521001119.
16. Wu Y-T, Zhang C-J, Mol BW, Kawai A, Li C, Chen L, et al. Early Prediction of Gestational Diabetes Mellitus in the Chinese Population via Advanced Machine Learning. The Journal of Clinical Endocrinology & Metabolism. 2021 Mar 8;106(3):e1191–205.
17. Lee S, Zhou J, Wong WT, Liu T, Wu WKK, Wong ICK, et al. Glycemic and lipid variability for predicting complications and mortality in diabetes mellitus using machine learning. BMC Endocr Disord. 2021 Dec;21(1):94.
18. da Silva DB, Schmidt D, da Costa CA, da Rosa Righi R, Eskofier B. DeepSigns: A predictive model based on Deep Learning for the early detection of patient health deterioration. Expert Systems with Applications. 2021 Mar;165:113905.
19. Howlett SE, Rutenberg AD, Rockwood K. The degree of frailty as a translational measure of health in aging. Nat Aging. 2021 Aug;1(8):651–65.
20. Gomez-Cabrero D, Walter S, Abugessaisa I, Miñambres-Herraiz R, Palomares LB, Butcher L, et al. A robust machine learning framework to identify signatures for frailty: a nested case-control study in four aging European cohorts. GeroScience. 2021 Jun;43(3):1317–29.
21. Aponte-Hao S, Wong ST, Thandi M, Ronksley P, McBrien K, Lee J, et al. Machine learning for identification of frailty in Canadian primary care practices. IJPDS [Internet]. 2021 Sep 10 [cited 2021 Dec 23];6(1). Available from: https://ijpds.org/article/view/1650
22. Ju C, Zhou J, Lee S, Tan MS, Liu T, Bazoukis G, et al. Derivation of an electronic frailty index for predicting short‐term mortality in heart failure: a machine learning approach. ESC Heart Failure. 2021 Aug;8(4):2837–45.
23. Park C, Mishra R, Sharafkhaneh A, Bryant MS, Nguyen C, Torres I, et al. Digital Biomarker Representing Frailty Phenotypes: The Use of Machine Learning and Sensor-Based Sit-to-Stand Test. Sensors. 2021 May 8;21(9):3258.
24. Zhavoronkov A, Mamoshina P. Deep Aging Clocks: The Emergence of AI-Based Biomarkers of Aging and Longevity. Trends in Pharmacological Sciences. 2019 Aug;40(8):546–9.
25. Raghu VK, Weiss J, Hoffmann U, Aerts HJWL, Lu MT. Deep Learning to Estimate Biological Age From Chest Radiographs. JACC: Cardiovascular Imaging. 2021 Nov;14(11):2226–36.
26. Sayed N, Huang Y, Nguyen K, Krejciova-Rajaniemi Z, Grawe AP, Gao T, et al. An inflammatory aging clock (iAge) based on deep learning tracks multimorbidity, immunosenescence, frailty and cardiovascular aging. Nat Aging. 2021 Jul;1(7):598–615.
27. Keane PA, Topol EJ. AI-facilitated health care requires education of clinicians. The Lancet. 2021 Apr;397(10281):1254.
28. Lenz S, Hess M, Binder H. Deep generative models in DataSHIELD. BMC Med Res Methodol. 2021 Dec;21(1):64.
Artificial intelligence (AI) as a concept was formulated already in 1956 by John McCarthy and can be formulated as computer algorithms mimicking features of human intelligence, such as learning and problem-solving. For the last decade, the tremendous growth of computational power and data availability paved the way for the widespread use of AI across all fields, and the medical field is no exception.
The current applications of AI are based on machine learning (ML) techniques – complex statistical models that can extract dependencies from the data without being explicitly programmed, make predictions, and some are able to learn new information. An example of the ground-breaking powers of AI is AlpaFold by DeepMind (1), which solved the fundamental problem of protein folding. ML provides an attractive alternative to other approaches in areas with little prior knowledge about those dependencies or where they are too complex. Current applications and successes of ML largely rely on available data volumes. The growth of data continues, and, as of today, the term big data refers to petabytes (1024 terabytes) and exabytes (1024 petabytes) of information.
The medical field is not excluded from an all-present spread of AI. The so-called digital health revolution resulted in the acquisition of massive clinical data in a diverse population. The volume and variety of data obtained through electronic health records and sensor-equipped devices give a vast field for the application of ML. Electronic health record systems allow to improve evidence-care guidance (2), and high-capacity and real-time data processing enhanced by ML give the possibility to provide personalized decisions (3). Also using AI at the bedside allows enhancing human-guided diagnostics and prediction. The Internet of Health Things (IoHT) concept includes smart health objects, data from which unites into one personal record (4). The application of ML together with IoHT may further improve the development of AI personalized systems.
A large emphasis is placed on improving the quality of care for older adults by implementing AI in general diagnostics and geriatric oncology programs (5). Though still in their early stages, such AI-based platforms hold promise for successful management of the complex conditions in the older population.
A variety of machine learning methods is widely used today, but the method that drives the most attention is the neural network, particularly, deep learning. Artificial neural networks (ANN) were initially created as a model inspired by a human brain, with an input layer that gets information and an output layer that produces a result. Each layer of ANN consists of nodes or neurons, each of which contains some non-linear processing function and possesses some weight. Early ANNs included only input and output layers and did not show a high predictive power. Introducing one or more hidden layers between input and output boosted ANN's predictive power, and modern deep neural networks (DNN) (6) include many hidden layers. Training a DNN revolves around the following:
The power of DNNs is their ability to learn complex abstractions by constructing meaningful features during the learning process (similarly to a human brain). The architecture of DNNs varies and can be incredibly complex, including different types of functions and different interconnections between layers. There are different subtypes of DNN, for example, the Recurrent Neural Networks (RNN), which have been applied in the health area because they are appropriate for the treatment of data such as text, speech, and DNA sequences (7). In this type of network, each state affects the results of the following one — thus presenting a memory in the network.
Despite the vast popularity of DNNs, other ML methods are still widely used and able to provide comparable results. Among the most powerful ones are the methods based on the decision trees – simple classifiers that can be depicted as a set of cascading decisions (if A then B, if not A then C) with multiple levels. Widely popular random forest and xgBoost techniques both use different ensembles of decision trees. In random forest decision trees are independent and model predictions are obtained by “voting”, while in xgBoost each new decision tree is trained to predict the error of the previous one thus minimizing the general error of prediction.
One of the most known applications of AI in diagnostics is IBM Watson Health™, which includes a range of ML models, including DNN ones, and promises high accuracy in the diagnosis of various diseases. IBM Watson learns on unstructured and semistructured data from the clinical literature, health records, and test results identifying the most important pieces of information and then mining a patient's data. The system then forms and tests hypotheses and finally provides a list of individualized recommendations, including a patient's eligibility for specific treatments. IBM Watson Health™ presently offers commercialized applications of the Watson system for genomics, drug discovery, health care management, and oncology (8). However, as with any AI, Watson is not performing equally well in all types of diagnostics. A study on the oncology patients in China showed that for ovarian cancer the concordance was 96% and for lung cancer and breast cancer obtained – slightly above 80%, but for colon cancer and cervical cancer – already lesser than 64%. And, for gastric cancer, the system reached a very low concordance of only 12% (9).
So, despite the hopes of obtaining a universal AI tool for all diseases, there is still a need for specialized AI systems. And such particular use various ML methods found in imaging (10); ophtalmology (11); diagnostic and outcome predictions for multiple types of cancer including breast (12,13) and skin cancer (14); cardiovascular diseases (15); and diabetes (16,17).
AI can also be used to assess aging and general health status not only through the identification and treatment of most widespread age-related diseases but also by detection of health deterioration, physical and cognitive frailty.
DeepSigns (18) is an RNN-based system that allows early detection of deterioration signs that uses the data of ICU (intense care unit) hospitalized patients. It uses the APACHE II index based on vital signs, such as blood pressure and temperature, and results of laboratory exams. Based on the patient's data, DeepSigns can predict future changes in vital signs with an accuracy of around 80%. The test results from DeepSigns identified 50% and 60% of otherwise unidentifiable cases. As 17% of the 50% and 60% tested patients from learning data died, it means that around 9% of deaths could be prevented with the DeepSigns model application to the early intervention in ICU units. The future application might include not only embedding DeepSigns in ICU warning systems but also in-home use for elderly high-risk patients.
Several AI systems showed promising results in addressing frailty - a geriatric condition linked to an elevated risk of rapid declines in health and function among the older population. The degree of frailty is commonly used as a measure of health in aging (19). Frailty depends on a multiplicity of factors, and ML was successfully used to analyze this complex multidimensional data. Gomez-Cabrero et al. (20) employed random forest classification to develop a frailty prediction system. Analysis of the models allowed them to identify both main protective biomarkers – vitamin D3, lutein zeaxanthin, and miR-125b-5p – as well as risk biomarker, cardiac troponin T. Another study by Aponte-Hao et al. (21) analyzed data on more than 5,000 non-frail and frail patients using a range of models including ANN and xgBoost. Authors were able to develop an accurate prediction system, and analysis of models revealed that frail patients were statistically significantly likely to be older, female, and less likely to have no known chronic conditions. Of those with at least one chronic condition, frail patients were more likely to have chronic obstructive pulmonary disease, dementia, depression, and hypertension. The advances in such ML models allow not only to extract patterns of frailty but to develop new indices able to predict short-term mortality as a part of AI integrated systems (22,23).
Aging clocks are the composite systems that can estimate an organism's biological age (age of its cells, tissues, and systems), capture different processes (e.g., cellular senescence) and consequences (e.g., risk of mortality). Such predictors of age based on deep learning are rapidly gaining popularity (24).
Currently developed aging clocks are based on a variety of data; for example, Vineet et al. developed an aging clock based on imaging data, biomarker data (C-reactive protein, glycated hemoglobin, albumin, total cholesterol, and others), physical activity, and anthropometry for more than 20 thousand individuals (25). For each type of data, they developed ML models using various types of neural networks, including DNN with different architectures and were able to build high accuracy predictors and estimate the dependencies between chosen parameters and health/disease status.
Another proposed aging clock, iAge, is based on inflammatory biomarkers (26). From the blood immunome data of more than 1,000 individuals from 8 to 96 years of age, the authors developed a deep learning model based on patterns of systemic age-related inflammation. The iAge is able to track immunosenescence, frailty, cardiovascular aging, and probability of exceptional longevity. Analysis of factors contributing to this model highlighted as one of the most important the chemokine CXCL9 involved in cardiac aging and poor vascular function.
Modern ML-based AI shows great promises in the medical field, particularly for tracking healthy aging, and potentially enables patients' access to the best quality of care. However, the digital revolution is not fast, and full implementation of AI requires both more technical advances and cross-training of clinicians and AI experts (27). The issue of preserving the privacy of patients' data is of great importance, and solutions are being developed (28). The potential inclusion of AI systems into healthcare will allow more precise and personalized treatment for more patients.
References
1. Thornton JM, Laskowski RA, Borkakoti N. AlphaFold heralds a data-driven revolution in biology and medicine. Nat Med. 2021 Oct;27(10):1666–9.
2. Hemingway H, Asselbergs FW, Danesh J, Dobson R, Maniadakis N, Maggioni A, et al. Big data from electronic health records for early and late translational cardiovascular research: challenges and potential. European Heart Journal. 2018 Apr 21;39(16):1481–95.
3. Rajkomar A, Dean J, Kohane I. Machine Learning in Medicine. N Engl J Med. 2019 Apr 4;380(14):1347–58.
4. da Costa CA, Pasluosta CF, Eskofier B, da Silva DB, da Rosa Righi R. Internet of Health Things: Toward intelligent vital signs monitoring in hospital wards. Artificial Intelligence in Medicine. 2018 Jul;89:61–9.
5. Extermann M, Brain E, Canin B, Cherian MN, Cheung K-L, de Glas N, et al. Priorities for the global advancement of care for older adults with cancer: an update of the International Society of Geriatric Oncology Priorities Initiative. The Lancet Oncology. 2021 Jan;22(1):e29–36.
6. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015 May 28;521(7553):436–44.
7. Ravi D, Wong C, Deligianni F, Berthelot M, Andreu-Perez J, Lo B, et al. Deep Learning for Health Informatics. IEEE J Biomed Health Inform. 2017 Jan;21(1):4–21.
8. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, et al. Artificial intelligence in healthcare: past, present and future. Stroke Vasc Neurol. 2017 Dec;2(4):230–43.
9. Zhou N, Zhang C, Lv H, Hao C, Li T, Zhu J, et al. Concordance Study Between IBM Watson for Oncology and Clinical Practice for Patients with Cancer in China. The Oncol. 2019 Jun;24(6):812–9.
10. Barragán-Montero A, Javaid U, Valdés G, Nguyen D, Desbordes P, Macq B, et al. Artificial intelligence and machine learning for medical imaging: A technology review. Physica Medica. 2021 Mar;83:242–56.
11. Hogarty DT, Mackey DA, Hewitt AW. Current state and future prospects of artificial intelligence in ophthalmology: a review: Artificial intelligence in ophthalmology. Clin Experiment Ophthalmol. 2019 Jan;47(1):128–39.
12. Houssein EH, Emam MM, Ali AA, Suganthan PN. Deep and machine learning techniques for medical imaging-based breast cancer: A comprehensive review. Expert Systems with Applications. 2021 Apr;167:114161.
13. Yu K, Tan L, Lin L, Cheng X, Yi Z, Sato T. Deep-Learning-Empowered Breast Cancer Auxiliary Diagnosis for 5GB Remote E-Health. IEEE Wireless Commun. 2021 Jun;28(3):54–61.
14. Khamparia A, Singh PK, Rani P, Samanta D, Khanna A, Bhushan B. An internet of health things‐driven deep learning framework for detection and classification of skin cancer using transfer learning. Trans Emerging Tel Tech [Internet]. 2021 Jul [cited 2021 Dec 23];32(7). Available from: https://onlinelibrary.wiley.com/doi/10.1002/ett.3963
15. Swathy M, Saruladha K. A comparative study of classification and prediction of Cardio-Vascular Diseases (CVD) using Machine Learning and Deep Learning techniques. ICT Express. 2021 Sep;S2405959521001119.
16. Wu Y-T, Zhang C-J, Mol BW, Kawai A, Li C, Chen L, et al. Early Prediction of Gestational Diabetes Mellitus in the Chinese Population via Advanced Machine Learning. The Journal of Clinical Endocrinology & Metabolism. 2021 Mar 8;106(3):e1191–205.
17. Lee S, Zhou J, Wong WT, Liu T, Wu WKK, Wong ICK, et al. Glycemic and lipid variability for predicting complications and mortality in diabetes mellitus using machine learning. BMC Endocr Disord. 2021 Dec;21(1):94.
18. da Silva DB, Schmidt D, da Costa CA, da Rosa Righi R, Eskofier B. DeepSigns: A predictive model based on Deep Learning for the early detection of patient health deterioration. Expert Systems with Applications. 2021 Mar;165:113905.
19. Howlett SE, Rutenberg AD, Rockwood K. The degree of frailty as a translational measure of health in aging. Nat Aging. 2021 Aug;1(8):651–65.
20. Gomez-Cabrero D, Walter S, Abugessaisa I, Miñambres-Herraiz R, Palomares LB, Butcher L, et al. A robust machine learning framework to identify signatures for frailty: a nested case-control study in four aging European cohorts. GeroScience. 2021 Jun;43(3):1317–29.
21. Aponte-Hao S, Wong ST, Thandi M, Ronksley P, McBrien K, Lee J, et al. Machine learning for identification of frailty in Canadian primary care practices. IJPDS [Internet]. 2021 Sep 10 [cited 2021 Dec 23];6(1). Available from: https://ijpds.org/article/view/1650
22. Ju C, Zhou J, Lee S, Tan MS, Liu T, Bazoukis G, et al. Derivation of an electronic frailty index for predicting short‐term mortality in heart failure: a machine learning approach. ESC Heart Failure. 2021 Aug;8(4):2837–45.
23. Park C, Mishra R, Sharafkhaneh A, Bryant MS, Nguyen C, Torres I, et al. Digital Biomarker Representing Frailty Phenotypes: The Use of Machine Learning and Sensor-Based Sit-to-Stand Test. Sensors. 2021 May 8;21(9):3258.
24. Zhavoronkov A, Mamoshina P. Deep Aging Clocks: The Emergence of AI-Based Biomarkers of Aging and Longevity. Trends in Pharmacological Sciences. 2019 Aug;40(8):546–9.
25. Raghu VK, Weiss J, Hoffmann U, Aerts HJWL, Lu MT. Deep Learning to Estimate Biological Age From Chest Radiographs. JACC: Cardiovascular Imaging. 2021 Nov;14(11):2226–36.
26. Sayed N, Huang Y, Nguyen K, Krejciova-Rajaniemi Z, Grawe AP, Gao T, et al. An inflammatory aging clock (iAge) based on deep learning tracks multimorbidity, immunosenescence, frailty and cardiovascular aging. Nat Aging. 2021 Jul;1(7):598–615.
27. Keane PA, Topol EJ. AI-facilitated health care requires education of clinicians. The Lancet. 2021 Apr;397(10281):1254.
28. Lenz S, Hess M, Binder H. Deep generative models in DataSHIELD. BMC Med Res Methodol. 2021 Dec;21(1):64.