Health care is gradually becoming an important field of artificial intelligence research and applications.
Today, almost every area of ​​the medical industry is affected by the rise of technology. For example, image recognition is revolutionizing the diagnostic process. Recently, Google's DeepMind neural network can diagnose 50 eye diseases that threaten vision, and the accuracy is comparable to medical experts. Even some pharmaceutical companies are trying to learn more about designing new drugs. For example, Merk worked with startup Atomwise, and GlaxoSmithKline and Insilico Medicine announced a partnership.
In the private market, healthcare AI ventures have received $4.3 billion in 576 financings since 2013, a figure that far exceeds other areas of artificial intelligence.
Artificial intelligence in the healthcare field now focuses on improving patient outcomes, adjusting the interests of various stakeholders, and reducing healthcare costs. One of the major obstacles to artificial intelligence in the healthcare arena is to overcome inertia, radically improve existing processes that are no longer effective, and try to apply emerging technologies.
Artificial intelligence faces the unique technical and viable challenges of the medical industry. For example, in the United States, patient data does not have a standard format and does not have a central repository. When patient files are sent by fax, mail, or handwritten pictures in an unreadable PDF format, extracting information from them will be a unique challenge for AI.
Large technology companies like Apple have their own advantages in this area, especially in the large network of partners joining healthcare providers and EHR (Electronic Health Record) providers.
Apple's design and development of ResearchKit and CareKit can generate new data sources and control EHR data in the hands of patients. These two software frameworks are expected to become revolutionary products for clinical research. In the first in-depth mining of industry AI, CB Insights data can be used to discover trends that are changing the healthcare industry.
AI is the rise of medical equipment
The FDA (US Food and Drug Administration) implements a rapid regulatory approval program for artificial intelligence software for clinical imaging and diagnostics. In April of this year, the FDA approved AI software for screening patients with diabetic retinopathy, which can accurately screen patients without the need for expert diagnostic advice. It has been given the title of “breakthrough deviceâ€, which speeds up the process of bringing products to market.
The software IDx-DR was able to correctly identify patients who “exceeded mild diabetic retinopathy†within 87.4% of the time, and determined those who did not have the disease within 89.5% of the time. IDx is one of the many software products approved by the FDA for clinical commercial applications in recent months.
Viz. Ai is approved for the analysis of CT scan images to detect stroke-related indicators and to inform patient staff of patient information in a timely manner. After obtaining FDA approval, Viz. Ai completed a $21 million Series A round of financing from Google Ventures and Kleiner Perkins Caufield & Byers.
GE Ventures' startup, Arterys, was approved by the FDA last year to analyze heart images through its cloud AI platform. This year, the FDA approved the approval of its AI software for liver and lung lesion location for cancer diagnosis. The rapid regulatory approval has opened up new business channels for more than 70 artificial intelligence imaging and diagnostic companies that have conducted equity financing since 2013, with a total of 119 financings.
The FDA is focused on clearly defining and managing "software, medical devices," especially considering the recent rapid development of artificial intelligence. FDA plans to apply the pre-cert program piloted in January this year to AI software.
The FDA added: "The program allows strange devices to make minor changes to their equipment and retrogrades without having to submit a certification application each time." FDA said that all aspects of its regulatory framework, such as software certification tools, will become "flexible enough" to accommodate The advancement of artificial intelligence.
Neural network found atypical risk factors
Using artificial intelligence, researchers began to study and measure atypical risk factors that were difficult to quantify in the past. Using neural networks to analyze retinal images and speech patterns helps identify people's risk of heart disease.
According to a paper published this year in Nature, Google researchers used a neural network trained to identify retinal images to identify cardiovascular risk factors. The study found that not only risk factors such as age, gender, and smoking patterns can be identified through retinal images, but also “quantitative precision that has never been seen beforeâ€.
In another study, the Mayo Clinic teamed up with Israeli founder Beyond Verbal, which focused on analyzing acoustic features in sound to identify the apparent acoustic characteristics of patients with coronary artery disease (CAD). . The study found that when the test subject describes the emotional experience, there are two sound features that are closely related to CAD.
A recent study by the founder Cardiogram showed: “After deep learning, the wearable heart rate sensor can detect changes in heart rate variability driven by diabetes after exposure to the human body.†The algorithm used by the sensor to measure diabetes by heart rate is as accurate as possible. 85%.
Artificial intelligence has the ability to detect disease patterns and will continue to pave the way for new diagnostic methods and identification of previously unknown risk factors.
Apple subverted clinical trials
Apple is building a clinical research ecosystem around devices such as the iPhone and Apple Watch. Data is at the heart of AI applications, and Apple can provide medical researchers with two previously difficult patient health data.
Although many companies strive to digitize their health records, achieving easy sharing of health information between organizations and software systems, or so-called interoperability, remains a major challenge in healthcare.
This problem is particularly evident in clinical trials, where accurately matching trials and patients is a time consuming and challenging process for both the clinical research team and the patient.
More than 18,000 clinical studies currently recruit patients only in the United States. If a doctor knows about an ongoing clinical trial, he or she will occasionally recommend it to his or her patient. Otherwise, it is only possible to pass a comprehensive federal database ClinicalTrials on closed and ongoing clinical trials. Gov recruits subjects.
Apple is trying to change the way information is transmitted in the healthcare sector and opens up new possibilities for AI, especially around how clinical researchers invest in and monitor patients.
Since 2015, Apple has launched two open source frameworks, ResearchKit and CareKit, to help clinical trials recruit patients and remotely monitor their health. These two frameworks allow researchers and developers to create medical applications to monitor the daily life of the subject.
For example, Duke researchers have developed an app, Autism & Beyond, which uses the iPhone's front-facing camera and facial recognition algorithms to screen out children with autism.
Similarly, approximately 10,000 users use the application mPower, which provides exercises such as finger tapping and gait analysis to determine if a patient has Parkinson's syndrome. These patients also agreed to share their data with the wider research community.
Apple is also working with EHR (Electronic Health Record) vendors such as Cerner and Epic to address interoperability issues. In January of this year, Apple announced that iPhone users can access participating organizations' electronic health records through the "health" application that comes with their mobile phones.
The "Health Record" feature is a derivative of the work of AI+ healthcare startup Gliimpse before it was acquired by Apple in 2016. The interface is simple and easy to operate, and users can easily find information about allergies, illnesses, immunizations, laboratory results, procedures, and vital signs.
In June, Apple introduced the Health Records API for developers. Users can choose to share data with third-party applications and medical researchers, which creates new opportunities for disease management and lifestyle monitoring.
Large pharmaceutical companies rebrand with AI
Nowadays, AI biotechnology startups are constantly emerging. Traditional pharmaceutical companies are experiencing unprecedented pressures and are turning their attention to AI+SaaS (software as a service) ventures, hoping to find innovative solutions.
In May of this year, Pfizer and XtalPi established a strategic partnership (XtalPi is an artificial intelligence startup company supported by technology giants such as Tencent and Google), hoping to predict the characteristics of small molecule drugs and develop "calculation-based" Rational drug design."
However, Pfizer is not the only one.
Top pharmaceutical companies such as Novartis, Sanofi, GlaxoSmithKlein, Amgen, and Merck have announced partnerships with AI Ventures in recent months to find new drug treatments for oncology and heart disease. a series of diseases.
The interest of pharmaceutical companies in this area has also contributed to the increase in the number of equity transactions, reaching 20 in the second quarter of 2018, equal to the total volume of transactions in 2017.
Although many AI+Saas startups are still in the early stages of investment, they have attracted many pharmaceutical companies to cooperate with them.
The application of AI in the medical industry is not limited to drug development. As one of the largest artificial intelligence M&A deals, Roche Holdings acquired Flatiron Health for $1.9 billion in February 2018. The latter can mine patient data through machine learning.
There are currently more than 2,500 clinics using Flatiron's oncology electronic medical record OncoEMR, and more than 2 million active medical records are available for research.
Roche wants to collect real world data (RWE) and analyze the sources of electronic medical records and other data to determine the benefits and risks of the drug. In addition to testing the safety of drugs after marketing, RWE can also help design better clinical trials and new treatments in the future.
AI needs a doctor
AI companies need medical experts to annotate images to teach algorithms how to identify anomalies. Technology giants and governments are investing heavily in this sector and opening up the database to researchers.
Google DeepMind teamed up with Moorfield's Eye Hospital two years ago to explore the use of AI for ocular disease detection. Recently, DeepMind's neural network was able to make correct referral decisions for 50 eye diseases that threaten vision, with an accuracy of 94%.
It is only the first stage of research. To train the algorithm, DeepMind puts a lot of time to mark and clean up the OCT (Optical Coherence Tomography) scan database for testing eye conditions and preparing for subsequent AI applications.
Alibaba also decided to apply AI to the diagnostic process around 2016.
According to Alibaba Cloud's chief AI scientist, Wan Lili, once the company collaborates with the medical structure to obtain medical impact data, it must hire experts to annotate image samples.
The AI ​​Unicorn Yitu Technology is trying to expand the field of artificial intelligence diagnosis. The company also emphasized the importance of the medical team in an interview with the South China Morning Post.
Yitu claims to have a team of 400 doctors to tag medical data, adding that because American doctors have higher salaries, American AI startups are also much more expensive to do. But in the United States, government agencies such as the National Institutes of Health (NIH) are studying artificial intelligence.
In July of this year, NIH released a database containing 32,000 lesions annotated and identified in CT images, which were provided anonymously by 4,400 patients. In addition, private companies such as Ventilation (GE) and Siemens are also looking for ways to create large-scale databases.
GE Healthcare received a patent in May this year that explored how to use machine learning to analyze cell types in microscope images. The patent proposes an "intuitive interface that allows medical personnel (such as pathologists, biologists) to annotate and evaluate different cell phenotypes used in algorithms and cell phenotypes presented through the interface.
Although some of the algorithms that have been proposed so far can be used to reduce manual processes, AI still relies heavily on medical experts for training.
China is developing rapidly
Chinese investors have begun to increase investment in overseas ventures, and local health care AI ventures are also growing. Chinese technology giants are bringing partnerships to bring products from other countries to the mainland.
In the past few years, China's trading activities have not been worth mentioning in the world, but now the ranking in the global healthcare AI market has risen sharply.
In the first half of 2018, China surpassed the United Kingdom and became the second most active health care AI transaction in the world.
After receiving $72 million in financing and gaining support from investors such as Sequoia Capital China, Infervision became the most funded Chinese venture in the AI ​​solution sector in China's healthcare industry.
At the same time, China's investment in foreign health care AI startups is also increasing.
Recently, Fosun Pharma bought a minority stake in the US Butterfly Network, Tencent invested in Atomwise, Lenovo invested in Lunit in South Korea, and IDG Capital invested in SigTuple in India.
The Chinese government released an artificial intelligence program last year with the goal of becoming a global leader in artificial intelligence research by 2030. Healthcare is one of the first four key areas of artificial intelligence applications in China.
China's focus on the healthcare industry is not just about becoming a global leader in AI technology.
According to last year's census, the long-standing one-child policy has led to an aging population: more than 158 million people over 65 years of age, coupled with a labor shortage, forcing the government to shift its focus to improving automation in the health care sector.
As early as 2016, China began to work hard to integrate medical data into a database. Similar to the United States, China also has problems such as data confusion and lack of interoperability.
To solve this problem, the Chinese government has successively opened a number of regional health data centers to integrate data on national insurance claims, birth and death registrations, and electronic health records. China's large technology companies are also supporting the health care AI field with the support of the government.
DIY diagnosis see here
Artificial intelligence is transforming smartphones and consumer wearables into powerful home diagnostic tools.
Venture company. Io claims to be trying to simplify the urine analysis step to make it as simple as taking a photo. Its first product, Dip. Io uses traditional urine analysis test strips to monitor a range of urinary tract infections. Computer vision algorithms can analyze test strips under different lighting conditions and camera quality with a smartphone.
It has been put into commercial use in Europe and Israel is Dip. Io recently received FDA approval.
In recent years, the penetration rate of smartphones in the United States has increased. At the same time, due to deep learning, the error rate of the image recognition algorithm is also significantly reduced. The combination of the two creates new possibilities for use as a diagnostic tool.
For example, SkinVision uses a smartphone camera to monitor skin lesions and assess skin cancer risk. The company received $7.6 million in financing from existing investors Leo Pharma and PHS Capital in July this year.
According to reports, the company's Amsterdam-based company will use the funds to promote FDA approval.
The new role of AI in value-based healthcare
Artificial intelligence began to play a role in quantifying the quality of services patients receive in hospitals.
A value-based service model focuses on patients, which motivates health care providers to deliver the highest quality care at the lowest possible cost.
This type of model contrasts with the service charging model in which service personnel are paid a certain percentage based on the number of services performed. The more procedures you have (for example, the more tests you have), the higher your financial rewards.
In 2010, the “Patient Protection and Price Medical Treatment Act†was passed, and the value-based service model began to enter the public eye.
Some of the existing safeguards include providing financial incentives to health care providers only in compliance with quality performance indicators, or penalties for hospital-acquired infections and preventable readmissions.
The goal toward a value-based healthcare system is to align the incentives of the service provider with the incentives of the patient and the payer. For example, under the new system, the hospital will give certain financial rewards to doctors who reduce unnecessary inspections.
Georgia-based startup Jvion works with suppliers such as Geisinger, Northwest Medical Specialties and Onslow Memorial Hospital.
Some of Jvion's case studies have highlighted the successful use of machines to identify patients who are at risk of re-admission within 30 days of hospitalization.
The nursing team can then teach the patient daily preventive measures based on Jvion's recommendations. The algorithm combines patient health data with historical data on socioeconomic factors (such as income, accessibility) and non-compliance to calculate risk.
Another method is for an insurance company to identify a patient at risk and to intervene by alerting the health care provider.
Artificial intelligence technology for in-hospital management solutions is still in its infancy, and startups are working to help healthcare providers reduce costs and improve care.
Treating robots with "can" and "can't"
Nowadays, mental health treatment costs are high and cannot be serviced around the clock. Many companies are turning their attention to the development of AI-based mental health robots.
Early ventures focused on the use of cognitive behavioral therapy, which gradually changed the patient's negative thoughts and behaviors. According to one approach, many emotional tracking and digital diary health applications on the market created dialogue expansion.
Woebot (clinical psychologist Alison Darcy developed a Facebook Messenger-based chat bot) received $8 million in financing from NEA, with a clear exemption statement that could not replace traditional therapy and human interaction.
Another company, Wyse, received $1.7 million in financing last year and launched an "anxiety and depression" robot on iTunes.
Startup X2 AI claims that its own AI robot Tess has more than 400 paying users. It developed a "belief-based" chat bot, "Sister Hope," to inform users of a clear disclaimer and privacy policy before starting a conversation (about messenger, chat is subject to Facebook's privacy policy, and session content is also visible).
But the accessibility of Fackbook, lack of regulations, etc. make it difficult to verify some robots and their privacy terms.
But as deep learning researcher Yoshua Bengio mentioned in a recent podcast about artificial intelligence elements, "AI is like an idiot scholar." There is no concept of psychology, what humans are, and how psychological changes work.
Mental health is a big point, and the symptoms and subjectivity are highly variable during the analysis. At the moment, artificial intelligence can do more than just regular inspections, but also foster a sense of “companion†with human language production.
However, as mentioned in a recent psychology article, our brain believes that we are chatting with robots without the complexity of deciphering nonverbal cues.
For more complex mental health problems, this can be problematic and likely to be dependent on robots and quick fixes, but in fact, these solutions do not have the ability to analyze or solve root causes in depth.
This is considered to be the most secure work in automation, requiring AI robots to have a high level of emotional cognition and freely interacting with each other. Despite the increased cost and accessibility, mental health care is still an extremely difficult task for AI.
Dehydrated onion powder is made of high-quality onions produced in green onion planting bases. It is processed by international leading technology. It is a pure natural product. This product is rich in nutrients such as a variety of vitamins, amino acids, and minerals. Appetite, lowering blood sugar, lipid-lowering blood pressure, anti-cancer and other effects, it is a multifunctional health food.
Dehydrated Onion Powder,Natural Plant Extract Onion Powder,Advanced Dehydrated Onion Powder,Delicious Onion Powder
Laian Xinshuyu Food Co., Ltd , https://www.xinshuyufood.com