How can artificial intelligence accelerate the study of many human diseases?

In !-- article, xiao compiled several important research results to explain how artificial intelligence can accelerate the research of many human diseases and share them with you! Photo Credit: CC0 Public Domain1 Nat Med: Developing a new AI diagnostic tool to predict the infection risk of COVID-19 without detection Doi:10.1038/s41591-020-0916-2 In a recent study published in the international journal Nature Medicine, scientists from King's College London and others developed an artificial intelligence diagnostic technique that can help predict an individual's likelihood of developing COVID-19 based on body symptoms.
This AI model can use data from the COVID Symptoms Research App to predict a person's risk of COVID-19 by comparing individual symptoms with the results of traditional COVID tests, which can help detect COVID-19 screening in restricted populations, and two clinical trials are currently under way in the UNITED Kingdom and the United States.
more than 3.3 million people worldwide have downloaded the app and used it to record their daily health, such as whether they feel well or have any new symptoms, such as persistent coughing, fever, fatigue and loss of taste. In
article, the researchers analyzed data regularly recorded in apps on their health, data from 2.5 million people in the United Kingdom and the United States, about a third of whom recorded symptoms associated with COVID-19, 18,374 individuals who said they had been tested for coronavirus, and 7,178 who tested positive.
(2) Neuro Oncology: AI technology helps brain cancer patients detect doi:10.1093/neuonc/noz199 A new study from Southwest University shows that by artificial intelligence, doctors can identify specific gene mutations in glioma tumors with more than 97 percent accuracy by simply examining three-dimensional images of the brain.
technology could replace the current practice of pre-treatment surgery, in which glioma samples are taken and analyzed to select the right treatment.
scientists across the country have been testing other imaging techniques in recent years, but the latest research may be one of the most accurate and feasible clinical methods.
"Know that a particular mutation state in a glioma is important for determining prognosis and treatment strategies, and the researchers say the ability to use traditional imaging and artificial intelligence alone to determine this state is a huge leap forward."
the study used deep learning networks and standard magnetic resonance imaging (MRI) to detect the state of a gene called isocric acid dehydrogenase (IDH), which produces enzymes that, in mutated states, can trigger tumor growth in the brain.
: AI predicts powerful antibiotics from more than 100 million molecules, killing super-resistant bacteria doi:10.1038/d41586-020-00018-3 In a study published in the international journal Cell, scientists from the Massachusetts Institute of Technology and others developed a ground-breaking machine learning method that identifies powerful new antibiotics from more than 100 million molecules, including one that can fight a variety of bacteria -- including tuberculosis and strains that are considered incurable.
researchers say the antibiotic, called halicin, is the first to be discovered by artificial intelligence.
although artificial intelligence has previously been used to assist in some parts of the antibiotic discovery process, they say this is the first time AI has identified a new class of antibiotics from scratch without using any human hypothesis.
researchers say the study is significant, and now the team has identified not only candidate molecules, but also promising ones in animal experiments.
, this method can also be applied to other types of drugs, such as drugs.
resistance to antibiotics has risen sharply around the world, and researchers predict that unless new drugs are urgently developed, drug-resistant infections could kill 10 million people a year by 2050.
but the discovery of new antibiotics and regulatory approvals have slowed over the past few decades.
people are constantly discovering the same molecules, we need new chemical reactions and new mechanisms of operation.
Rev Neurol: Artificial intelligence technology may accelerate and improve the diagnosis of Alzheimer's disease doi:10.1038/s41582-020-0377-8 Recently, a study published in the international journal Nature Reviews Neurology In the study, scientists from institutions such as the University of Sheffield said the use of artificial intelligence (AI) could help quickly diagnose Alzheimer's disease and improve the prognosis of patients; in the paper, researchers analysed how AI could be used in the medical field to help improve the time and economic impact of common neurodegenerative diseases on the NHS, such as Alzheimer's and Parkinson's disease.
The main risk factor for many neurological disorders is age, and as life expectancy increases globally, the number of people with neurodegenerative diseases is expected to reach an unprecedented number, with researchers predicting that the number of Alzheimer's patients alone will triple by 2050. Reaching 115 million, this poses a challenge to health systems; in this study, researchers used AI techniques, such as machine learning algorithms, to detect neurodegenerative diseases before symptoms worsened, improving patients' chances of benefiting from successful disease modification therapies.
: Artificial intelligence helps predict drug-resistant superbug doi:10.1073/pnas.2008807117 In a recent study, biomedical engineers at Duke University have shown that Machine learning analyses of growth dynamics to distinguish between different strains can then accurately predict other characteristics, such as resistance to antibiotics; the technology has the advantage of identifying diseases and predicting strain behavior more accurately than current standard techniques, the results of which are published in the journal PNAS.
!--/ewebeditor:page--!--ewebeditor:page title"--In the history of microbiology, bacterial identification relies on growing cultures and analysis of the physical properties and behavior of the resulting bacterial bacterios.
until recently, scientists were able to simply conduct genetic tests.
, gene sequencing is not universal and usually takes a long time.
even with the ability to sequence the entire genome, it may be difficult to link specific genetic variations to different behavioral characteristics.
photo Source: Wikipedia:6 Nat Metabol: New algorithms or artificial intelligence technology to help manage doi:10.1038/s42255-020-0212 In a recent study published in the international journal Nature Metabolism, scientists from oregon Health and Science University and others used artificial intelligence and automated monitoring to develop a new method that could help people with type 1 diabetes better manage their blood sugar levels. 'Our system design is unique,' said
researcher Dr. Nicole Tyler. 'We used special mathematical simulators to design artificial intelligence algorithms, but when they were validated on real data from people with type 1 diabetes, the recommendations were highly similar to 3d recommendations made by endocrinologists.'
this is important because it often takes 3-6 months for diabetics to make appointments and see an endocrinologist.
In this time, if the body's blood sugar levels are too high or too low, there is a risk of risky complications, and people with type 1 diabetes themselves cannot produce enough insulin to control their blood sugar by using insulin pumps or multiple insulin injections a day, a new algorithm developed by researchers that uses data collected from continuous glucose monitoring devices and wireless insulin pens to guide patients' treatment.
when used with a mobile app called DailyDose, the algorithm's recommendations showed that it was consistent with doctors 67.9 percent of the time.
(7) Cell: Chinese scientists have developed an artificial intelligence system that accurately diagnoses new coronary pneumonia and evaluates prognosis doi:10.1016/j.cell.2020.04.045 Recently, AI has made exciting new advances in many medical applications, which have inspired innovative developments in new AI-based radio diagnostic technologies.
Chen et al. reviewed various quantitative models of chest thin-layer CT and showed the effectiveness of quantitative tools in precise diagnosis and longitudinal follow-up.
study showed that deep learning algorithms help identify head CT scan abnormalities and assist in clinical misdiagnosis.
recent studies have demonstrated the potential of integrating AI into ophthalmology and childhood disease diagnosis systems, and have found that this can significantly improve the efficiency and accuracy of clinical diagnosis.
With the more accurate CT scanning tools, in a new study, the University of Science and Technology of Macau, Sichuan University Huaxi Hospital, Guangzhou Regenerative Medicine and Health Guangdong Province Laboratory, Tsinghua University, Zhongshan University, Three Gorges University, Anhui Medical University, Wuhan University, Guangzhou Medical University, Yunnan Province, the first Researchers at People's Hospital, the Hong Kong Polytechnic University and Guangzhou Kangrui AI Technology hypothesically believe that an AI system that accurately diagnoses NCP will help radiologists and clinicians manage patients who are prompted to develop COVID-19 NCP symptoms.
findings were published online in the Cell journal in the form of manuscripts of papers.
Another pressing need is to identify patients at higher risk of acute respiratory failure so that they can be closely monitored earlier and treated with early intervention, otherwise they will end up with a higher risk of multiple organ failure, accompanied by a higher mortality rate.
Given that the overall characteristics of lesions, including the number, size and density of lesions, and the overall state of the lungs are indicators of lung damage and remaining lung reserves, the authors also tested the hypothesis that clinical data and CT parameters can be used to build an AI system to produce accurate clinical prognostic models that will enable clinicians to develop plans for early monitoring and management of such patients.
, the authors constructed a large CT data set for NCP, other common pneumonia, and normal control groups, and established an AI diagnostic system to aid accurate diagnosis in one epidemic area and two non-epidemic areas in China.
: Diagnosing Breast Cancer Artificial Intelligence Wins Over Human Experts! Doi:10.1038/s41586-019-1799-6 In a recent study published in the international journal Nature, scientists from Google Health developed a new computer program that can diagnose and detect breast cancer with greater accuracy than human experts.
Breast cancer is the most common type of cancer in the female population, with more than 2 million newly diagnosed cases last year alone, and regular screening is essential to detect early symptoms of the disease in the group of patients with no obvious symptoms;
But there is often room for error in the interpretation of scan results, and a small percentage of all mammoth X-rays show false positives (misdiagnosing healthy people as having cancer) or false negatives (misdiagnosing disease positives as negative).
In this study, researchers used artificial intelligence models to scan thousands of women in the UK and the US for breast cancer; these images have been analysed and examined by doctors in real life, but unlike in clinical environments, machines (AI algorithms) do not diagnose diseases based on a patient's medical history.
9 ERMD: Breakthrough! Scientists are expected to develop a new type of breast cancer artificial intelligence diagnostic tool doi:10.1080/14737159.2019.1659727 !--/ewebeditor:page--!--ewebeditor:page In a study published in the international journal Expert Review of Molecular Diagnostics, scientists from Lancaster University and others have developed a new method that could identify specific chemical "fingerprints" of different types of breast cancer that could be used to develop AI software to create a new tool to diagnose breast cancer quickly and accurately.
article, researchers analyzed living tissue using a specialized chemical analysis technique called raman spectroscopy to identify the molecular structural properties of multiple types of breast cancer and the differences between each type of cancer.
spectroscopic analysis can provide real-time information about cells and can be used to detect cell behavior, diffusion, and when they appear in the body.
when chemical fingerprints of breast cancer cells are identified and how they change, researchers can use this information to train sophisticated machine learning algorithms to identify four different cancer subtypes.
10) Nat Machine Intell: Artificial Intelligence Power Biomedical Imaging Doi:10.1038/s42256-019-0095-3 According to a recent study, the Federal Institute of Technology zurich and .