Big data, AI, and ML (Machine Learning) have never been so hot, and their benefits have benefited companies in almost every industry.
Now is the time for companies to take advantage of these benefits. Before the value of AI and ML can benefit companies, companies must thoroughly understand their role in the trade business, the problems that can be solved, and the way to agree with corporate goals or expected results. For example, Google is already solving large-scale health problems. Bet.
Tech giant invests in AI healthcare
Google AI has been conducting AI-related research and collaboration projects in the fields of healthcare and biosciences, and it claims:
"There are dozens of potential application areas for machine learning, but healthcare is an extraordinary opportunity for the benefit of the people, and Google AI works closely with clinicians and medical service providers, hoping to develop tools to greatly improve medical services Availability and accuracy. "
Ron Moody, chief medical officer of Accenture Consulting, wrote a related article entitled "AI in Healthcare-A Key to Industry Evolution", The text reads:
"Artificial intelligence (AI) will revolutionize health care operations, health research, medical security delivery, and helping patients stay healthy. It can now be seen that this will be a key part of a long-term health care strategy."
AI has been integrated into the industrial processes, applications and systems that people deal with every day, which has helped the AI field to achieve further AI expansion. Using AI to support a series of activities, such as detecting diseases and medical diagnosis, will make a breakthrough and surpass in this field.
Healthcare has a similar picture with other commercial industries, administration, logistics, business processes, and customer relationships. In the above fields, the application of AI has begun to reduce costs and improve efficiency. The need for medical health to reduce costs puts tremendous pressure on the field. In these areas, institutions should now explore, invest, and use AI to achieve change.
According to Caserta's article, "The United States spends over $ 10,000per capita, or 18% of its GDP, on healthcare" ), And according to data from BGV (Benhamou Global Ventures), due to an aging population, an expanding market, and increasing labor costs, global healthcare costs are expected to increase at a rate of 4.1% per year from 2017 to 2021. In addition to digital disruption, innovative emerging companies have unique opportunities to rise, and use the healthcare ecosystem to build technology and solve specific problems.
The value of AI
By applying AI to business, logistics, administrative processes, and even using it to improve customer engagement, some medical and health institutions have seen the benefits of AI, but AI has the potential to continue expanding areas such as pathology and radiology. Interpretation and other fields. As data appears faster and faster, and the number of patient data sources continues to expand, AI can provide better support in data processing, visualization, and decision making.
In the current health care industry, AI can bring great value and better results with the help of ML and natural language processing (NLP). The technology used in health care will also support the new model of "value-based care." And with the increase of big data, the use of technology has a leverage effect, which makes the services for patients in medical care more personalized, and can promote transformation. The growth of the AI health market is expected to reach US $ 6.6 billion in 2021, a compound annual growth rate of 40% (see Figure 1).
Use cases for deep learning and computer vision
The field of computer vision has made great progress with the help of an AI technology called deep learning, or deep neural network. Taking full advantage of this AI technology in pocket technology products is like a scene from a science fiction movie ten years ago.
If these emerging and advanced computer vision systems can accurately classify cars or different kinds of dogs in the picture, then Google's engineers and scientists have also begun to think, "Can these systems learn to identify diseases presented in medical images? In a post called "Deep Learning for Detection of Diabetic Eye Disease" published by Google's AI blog, Google's product manager and double doctor of medicine Lily Peng and Research engineer Dr. Varun Gulshan described their exploration in the field of ophthalmology using computer-assisted diagnostic screening and detected an eye disease called diabetic retinopathy. Diabetic retinopathy is the fastest growing cause of preventable blindness worldwide. Normally, a trained doctor will be able to do this by examining a retinal scan of the patient's eye.
In the field of digital pathology, Google is developing deep learning algorithms that may help pathologists detect breast cancer in lymph node biopsies.
Improve diagnosis
Relying on AI and ML, the diagnostic accuracy of drugs has greatly improved in recent years. A current survey by the Big Market Research shows that the AI smart medical market will exceed $ 18.12 billion by 2025. An article about using AI to maintain health also showed that compared with humans, AI and ML can detect abnormalities faster and more accurately in scanning.
The use of AI and ML projects to detect abnormal phenomena that may be overlooked by human eyes is continuously improving the diagnosis and providing better care for patients. Here are a few examples:
Researchers at Stanford University have developed algorithms to interpret chest X-rays. This interpretation is not only as accurate as the radiologist's, but also takes a short time.
Doctors use Viz.ai's technology to save valuable diagnostic time by using technology to quickly and accurately detect blood clots in stroke patients before major injuries occur.
lower the cost
In the United States, 25% of health care costs are administrative costs, which far exceeds other developed countries. An important area where AI can make a significant impact is medical coding and medical billing, and AI can develop automated methods in this area.
Forbes has published an article titled AI And Healthcare: A Giant Opportunity, in which data show that the key to solving cost structure issues in current healthcare systems is time-consuming manual tasks Handing over to a machine allows patients to self-service their care needs anywhere. Human labor is to make more people live healthier lives, and self-service can reduce the labor required. According to a report from Accenture, by 2026, major clinical health AI applications could save the US healthcare market $ 150 billion annually.
The figure below shows the top ten health care AI projects that have saved a total of $ 150 billion. Accenture's assessment defines the impact of each application, the likelihood of adoption, and its value to the healthcare market. The top three applications represent the largest near-term value: robot-assisted surgery ($ 40 billion), virtual nursing assistant ($ 20 billion), and administrative workflow assistance ($ 18 billion).
Detecting medical fraud
Splunk's report "Artificial Intelligence and Machine Learning in the Regulatory Industry" shows that it is extremely difficult and time-consuming to detect or detect abnormal behavior patterns in a large number of healthcare providers (sometimes it takes months or even years to complete) . In addition, finding and closely monitoring the source of illegal prescriptions is another challenge. Medical fraud and drug abuse are endless, such as the large-scale fatal drug abuse in the United States due to the prescription and distribution of opioids. Machine learning can help find anomalous and potentially fraudulent vendors, which is difficult and time-consuming for humans.
Aggregating claims data can give people a complete picture of opioid purchases. Healthcare companies can use machine learning programs and algorithms to see models in the data and where the data deviates from the model. As a result, healthcare organizations can turn to preventing and detecting fraud rather than "pay chasing." According to an article by PBS News, the U.S. Department of Justice states that "almost 13 million illegal opioids in the United States are suspected of fraud and false billing," involving 23 pharmacists and 19 nurses. The types of data sources include: electronic health records (EHR), health level 7 (HL7) messages, medical devices, desktops, servers, storage devices, networks, portals, billing systems, and patient management systems.
new medical research
According to a Medium article by Daphne Koller, a MIT researcher and founder of startup Insitro, large pharmaceutical companies have been struggling to develop new treatments. Over the past few decades, drug development has become increasingly difficult and expensive to meet the needs of many patients. Drug clinical trial success rates hover in the single digits; pre-tax R & D costs for the development of new drugs (once they fail) are expected to exceed $ 2.5 billion; the return on drug development investment 30 years ago was $ 200 million, and the rate of return has declined linearly year after year ; According to some estimates, this number will drop to 0% by 2020.
Regulatory oversight and smaller patient databases increase costs. Ke's company is trying to change that, and Insitro is trying to revolutionize the pharmaceutical development process by using machine learning for drug development and treatment. Insitro has raised more than $ 100 million from well-known investors such as ARCH, Foreste Capital, Andreessen Horowitz, and Jeff Bezos Personal Ventures in a few months. On April 16, 2019, pharmaceutical giant Jared Scientific said it will pay Insitro $ 15 million, and if it reaches its stated goal, it will pay another $ 1 billion to develop a drug to treat a common liver disease, a liver disease It is called non-alcoholic steatohepatitis (NASH). The disease is rapidly becoming a global epidemic due to poor diet and lack of exercise. The cooperative agreement requires Insitro to create disease models, find the right targets for treating the disease, and test whether artificial intelligence can help drug development.
Personal Care
The Human Genome Project (HGP) 's first human genome sequencing took almost 13 years and cost £ 2.7 billion. Since then, technological advances have significantly reduced the time and cost of sequencing a single genome. Healthcare services begin using genome sequencing, reading patient data, and tailoring and optimizing care for patients based on their unique genetic characteristics. In addition to advances in genomic sequencing technology, the widespread application of big data and cloud technologies in the medical field has also made new advances in precision medicine.
Machine learning algorithms use cloud computing data lakes and data warehouses to identify patterns and make predictions. These data lakes and data warehouses can clean up (create a single "true source" in the data) and store large amounts of data to unify multiple healthcare systems. In this way, electronic health records can provide people with better and more targeted care.
Oncology and cancer research invest heavily in precision medicine by studying cancer genetics. In some cases, cancer may be genetically driven rather than its physical location in the patient. The Moffitt Cancer Center in Tampa, Florida has been working to integrate molecular genomics, demographics, and trial results to model each patient.
Sometimes incompatible treatments and medicines have risks. Precision medicine can not only reduce the risks, but also provide new solutions to fight diseases and provide medical services. A startup called Deep Genomics uses artificial intelligence and the genome to find the best drug therapy for every patient.
More and more healthcare organizations are investing, experimenting and integrating emerging technologies into their systems, which will break and change the traditional healthcare model.
in conclusion
Despite numerous use cases and evidence that the benefits of using artificial intelligence and machine learning in the healthcare industry are considerable, and that more and more venture capital flows into the industry each year, healthcare and technology startups still face a range of challenges The challenges faced by the huge potential market are not solvable only through technological improvements. Some of these challenges include:
Achieve wider use
· Implementation of tools
· Healthcare ecosystem
· Regulations
· Current business model
· Incentive alliances with payer-payee relationships
Healthcare organizations must learn to trust and use algorithms, which means that healthcare organizations want to see clinical validation of algorithms. To date, some people remain cautious or hesitant to fully adopt AI tools without substantial evidence to validate the results.
A digital health survey conducted by Accenture shows that skepticism is not just about doctors: about 25% of respondents are not yet ready for AI-powered health services. Many people are worried about the technology, such as they don't understand how artificial intelligence works, or they worry that the technology will really "read" them.
Another challenge relates to scale. Often within a limited scope, a proof-of-concept (POC) or pilot project that is tested may not be ready for rollout in a large institution. Another problem is that current advanced technology solutions may be too expensive for smaller regional and rural medical institutions in the early days.
Finally, consider that the current health industry involves large and diverse stakeholders. The impact of artificial intelligence on health care operations and treatment of medical managers, clinical practitioners, and patients will have some driving effects. Morgan Stanley estimates that the outlook is still good. The global artificial intelligence market in the medical field will soar from $ 1.3 billion today to $ 10 billion in 2024, with a compound annual growth rate of 40%.
Let's share the dry goods of AI learning and development together
Welcome to the whole platform AI vertical media self-reading
0 Comments