Artificial Intelligence (AI) is already driving major changes in the way things work and in the way people and companies relate to technology. In a much faster, intuitive and intelligent way, computers are analyzing data and seeking answers to the most different problems and questions.
In the area of health is no different . Artificial Intelligence in medicine has been positively impacting and is a great aid to increase the capacity of patient care and to act in the prevention and treatment of diseases.
In this article, we will bring some concepts about Artificial Intelligence, such as Machine and Deep Learning , and how the cognitive ability of computers is being applied in medicine. O TensorFlow , an AI software library created by Google, facilitates the existence of these applications and is already used in telemedicine.
IA x Machine Learning X Deep Learning
The definition of Artificial Intelligence is related to the ability of machines to imitate behavior and think as human beings – from having the ability to perform simple tasks in an automated way to a more complex cycle, such as perceiving, learning, reasoning and rational decision making. intelligent. This technology has a number of applications, such as popular voice assistants (Siri, Google Assistant, Alexa and Bixby), Facebook face recognition, Google autocomplete search, and Waze route suggestions.
THE IA’s popularity exploded a few years ago , largely due to the availability of GPUs ( graphic processing units ), which make parallel processing faster, cheaper, and more powerful than it was decades ago. It also has to do with the flood of data we have on the internet today (the big data ) with images, texts, transactions, map data, various personal information, which are made available on the net daily by all of us.
Already Machine Learning , Deep Learning and the ‘new’ Natural Language Processing are the ways you are teaching computers to think. These terms make up the AI:
- Machine Learning – Machine learning is the main driver of artificial intelligence. You can define Machine Learning such as the use of algorithms in computers, so that they collect data and learn from them, that is, the machines gather the information they receive or search the network, cross everything, and arrive at a result autonomously and much faster than a human would be able. One example is custom recommendations on Netflix and Amazon. So instead of someone implementing software routines “on hand,” giving instructions for the computer to complete a particular task, the machine is “trained” to perform the task on its own.
- Deep Learning – O Deep Learning uses complex algorithms to mimic the neural network of the human brain and to learn an area of knowledge. Car driving alone, predictive health, more pertinent recommendations, advanced facial recognition and voice are some examples of the application of Deep Learning , technology widely used by companies in various business segments .
- Natural Language Processing (PLN) – is the ability of machines to understand the language of human beings. Uses techniques to find patterns and recognize how people communicate in written and spoken form. PLN can be applied in social network post analysis to understand how customers feel about specific brands or products. It is also already used in bots that respond to queries made in the Facebook chat of companies .
The applications of Artificial Intelligence in medicine
And how is Artificial Intelligence being used in medicine? It is a fact that the large volume of information provided by medical systems, processed by new ways of learning machines, drives health care, helps reduce costs and improves process quality, accuracy of diagnoses and treatments.
An important function of Artificial Intelligence in medicine is to aid in the diagnosis of diseases. Google’s TensorFlow, for example, was used by a system that helps identify diabetic retinopathy , a major cause of blindness among adults. The program uses Deep Learning , analyzing retinal photographs to find similarities. In tests, the tool was able to achieve the same success rate of doctors.
O Watson, algorithm developed by IBM is another example that helps in the treatment of cancer. The tool also uses deep learning and based on evidence in the scientific literature and clinical and genetic data of the patient, indicates possible treatments. The computer does not say which is the best but brings all the cancer treatments to the case and its scientific evidence, including with degree of risk and side effects.
At the Albert Einstein Hospital in São Paulo, there are imaging devices capable of pointing out possible diseases and automatically forward notifications to the physician, and equipment that sends vital patient signals directly to the medical records.
TensorFlow: Google’s telemedicine algorithm
In tests passed via telemedicine it is possible to use machine learning to do the sorting and put the urgencies first in the analysis queue. From the reading he has done of other exams, the computer assists in the issuance of more accurate reports. In this case, a very efficient tool is the TensorFlow , an artificial intelligence software library created by Google.
In telemedicine systems, the solution is used as visual recognition to filter out exams with critical diagnostic possibilities, and by analyzing the data and images of the exam, artificial intelligence learns and can distinguish urgencies in more critical exams assisting physicians. If the suggestions do not make sense, the doctor can inform the system, which uses this information to automatically improve itself.
With the text and image analysis tools provided by the Google tool it is also possible to report almost immediately whether the quality of the exam is compromised and whether it will need to be redone. This is especially useful since the doctor does not “waste time” with a test that will not provide relevant information. And it prevents the reconvocation of patients to repeat the test.
THE Telemedicine Portal , for example, already applied the TensorFlow in several images. The Artificial Intelligence system is capable of analyzing data and detecting anomalies in the exams, helping doctors to bring more accurate diagnoses.
How TensorFlow Works
TensorFlow is a highly scalable machine learning which can run on a simple smartphone, computer or data center. The platform is open source and all specialized algorithms for solving specific questions are made available in the cloud through APIs.
With this format, the solution allows to build and train neural networks and quickly generate a product or service from the predictive model trained. Therefore, their applications are numerous and can be used by individuals searching for research or even large companies that need to implement strategies of Artificial Intelligence.
To learn more about Artificial Intelligence and to continue to update your knowledge about health technology, follow the Telemedicine Portal blog . You can also receive the articles by email when you subscribe to our newsletter, direct on site .