March 25, 2019 at 11:41Cancer, autism and diabetes: what AI will help treat in 2019
The AI AI market in medicine, according to Frost & Sullivan forecast, will reach $ 1.7 billion. By 2021, it will grow another three times - up to 6.6 billion. About which diseases will be treated with the help of artificial intelligence and which companies plan to capitalize on this, in a review of the Binary District.
A neural network that knows everything about cancer
“Watson, is there such a difficult case with me that you would advise?” - probably this is how doctors could turn to the artificial intelligence invented by IBM. The IBM Watson Supercomputer was created in 2011 and named after IBM's first president, Thomas Watson. In 2014, the company invested a billion dollars in the development of this project. Now one of its modules, Watson for Oncology , is used to diagnose and treat cancer.
The main task of Watson (the computer, not the president) is to understand what he is told and look for the necessary information in natural language in the databases. In the case of Watson for Oncology, this database includes more than 600 thousand medical reports and diagnoses, as well as two million pages of texts from medical journals and clinical trials in the field of oncology.
Based on these data, the AI and medical records of a particular patient make a diagnosis and tell doctors how best to treat cancer in specific cases. Information about the patient enters the computer in a depersonalized form.
A neural network can offer several treatment options, the doctor will have to choose the best one.
The attending physician can add information to the system. For example, write: “the patient has blood in the sputum when expectorating” - the computer will understand what is said and after 30 seconds it will issue an updated diagnosis and an updated course of treatment.
In 2016, AI identified a rare form of leukemia in a 60-year-old patient who was initially diagnosed with the wrong diagnosis. To do this, the system in 20 minutes studied 20 million scientific articles on cancer.
Starting in 2019, IBM Watson for Oncology will be used to treat cancer in US Army veterans. Other Watson modules are used to diagnose other diseases. By 2021, Frost & Sullivan predicts IBM will occupy 45% of the medical AI market, which by then will be $ 6.6 billion.
Last May, it was reported that Sberbank will use IBM Watson in its insurance system. Then “Sberbank life insurance” (the “daughter” of Sberbank) concluded the first contract in Russia with IBM for the use of the Watson for Oncology system.
An example of a more traditional type of AI in medicine is QTrobot: a robot for treating children with autism spectrum disorders. It is difficult for such children to communicate with other people: they poorly recognize other people's emotions and express their own. With age, this problem worsens. If you do not pay attention to her already in the first years of a child’s life, later it will be much more difficult for him to cope with this.
QTrobot is intended for children over the age of four. He communicates with them with words, gestures and facial expressions. Before using it, you can easily program it yourself - according to the founder, even a parent without technical skills can do it in less than 20 minutes. How to do it - shownin two short videos on the project website. With the help of a robot, you can teach a child to understand someone else’s mood, communicate and perform simple tasks - as, for example, in this video .
The startup LuxAI, which developed the robot, was created by researchers from the University of Luxembourg in 2016. The final prototype of the robot was presented a year later, and the first QTrobot test was carried out in 2018: scientists compared the treatment session of fifteen boys with autism from four to 14 years old with a robot and a doctor. It turned out that children pay more attention to the robot: on average, they looked at it twice as long.
The only drawback of QTrobot, like many other robots of this type, is its price: in 2017, the developers said that it would be 5500 euros. While the robot, however, did not go on sale: according to the scientific portal IEEE Spectrum, it is undergoing test tests at medical centers in Luxembourg, France, Belgium and Germany.
JARVIS (Just Another Very Intelligent System) is the voice platform interface for diabetics Diabnext AI . Named after the butler of Iron Man, Tony Stark, it complements the CLIPSULIN system developed by the company. The system includes “smart” trackers that are mounted on an insulin pen and on a glucometer (a device for measuring blood sugar), a mobile application and a computer program for storing and analyzing records. In 2017, CLIPSULIN won the 2017 CES Innovation Award in the biotechnology category.
The system helps a diabetic to keep a daily diary of his condition: record insulin doses, meals eaten, and other parameters. A diary is needed for diabetics to control their blood sugar. Typically, entries are made into it manually - in special notebooks or in mobile applications. After each meal, the diabetic records the approximate amount of carbohydrates contained in it and the dose of insulin set for them. CLIPSULIN trackers allow you to do this automatically.
The principle of operation of the device is described in the video presented in the company's blog : a diabetic takes pictures of breakfast, and the image recognition system built into the tracker determines how much carbohydrate it contains. After that, the patient sets a dose on his insulin pen: the number of units is immediately fixed.
Another tracker is connected to the meter: using it, you can transfer data about each measurement of sugar to a mobile application and to a computer. You can also enter other parameters into the system: the results of standard tests, the level of glycated hemoglobin, data on weight, exercises, and so on.
At the same time, Jarvis can answer user questions and, like any neural network, it learns in the process of data collection. As the developers explain , if a person, for example, periodically eats a lot of pizza, AI will analyze how this type of food affects blood glucose levels and helps to calculate the dose of insulin more accurately. A doctor who has a persistent diabetic can also access Diabnext if the patient so desires.
A neural network that knows everything about cancer
“Watson, is there such a difficult case with me that you would advise?” - probably this is how doctors could turn to the artificial intelligence invented by IBM. The IBM Watson Supercomputer was created in 2011 and named after IBM's first president, Thomas Watson. In 2014, the company invested a billion dollars in the development of this project. Now one of its modules, Watson for Oncology , is used to diagnose and treat cancer.
The main task of Watson (the computer, not the president) is to understand what he is told and look for the necessary information in natural language in the databases. In the case of Watson for Oncology, this database includes more than 600 thousand medical reports and diagnoses, as well as two million pages of texts from medical journals and clinical trials in the field of oncology.
Based on these data, the AI and medical records of a particular patient make a diagnosis and tell doctors how best to treat cancer in specific cases. Information about the patient enters the computer in a depersonalized form.
A neural network can offer several treatment options, the doctor will have to choose the best one.
The attending physician can add information to the system. For example, write: “the patient has blood in the sputum when expectorating” - the computer will understand what is said and after 30 seconds it will issue an updated diagnosis and an updated course of treatment.
In 2016, AI identified a rare form of leukemia in a 60-year-old patient who was initially diagnosed with the wrong diagnosis. To do this, the system in 20 minutes studied 20 million scientific articles on cancer.
Starting in 2019, IBM Watson for Oncology will be used to treat cancer in US Army veterans. Other Watson modules are used to diagnose other diseases. By 2021, Frost & Sullivan predicts IBM will occupy 45% of the medical AI market, which by then will be $ 6.6 billion.
Last May, it was reported that Sberbank will use IBM Watson in its insurance system. Then “Sberbank life insurance” (the “daughter” of Sberbank) concluded the first contract in Russia with IBM for the use of the Watson for Oncology system.
A robot that teaches children with autism to smile
An example of a more traditional type of AI in medicine is QTrobot: a robot for treating children with autism spectrum disorders. It is difficult for such children to communicate with other people: they poorly recognize other people's emotions and express their own. With age, this problem worsens. If you do not pay attention to her already in the first years of a child’s life, later it will be much more difficult for him to cope with this.
QTrobot is intended for children over the age of four. He communicates with them with words, gestures and facial expressions. Before using it, you can easily program it yourself - according to the founder, even a parent without technical skills can do it in less than 20 minutes. How to do it - shownin two short videos on the project website. With the help of a robot, you can teach a child to understand someone else’s mood, communicate and perform simple tasks - as, for example, in this video .
The startup LuxAI, which developed the robot, was created by researchers from the University of Luxembourg in 2016. The final prototype of the robot was presented a year later, and the first QTrobot test was carried out in 2018: scientists compared the treatment session of fifteen boys with autism from four to 14 years old with a robot and a doctor. It turned out that children pay more attention to the robot: on average, they looked at it twice as long.
The only drawback of QTrobot, like many other robots of this type, is its price: in 2017, the developers said that it would be 5500 euros. While the robot, however, did not go on sale: according to the scientific portal IEEE Spectrum, it is undergoing test tests at medical centers in Luxembourg, France, Belgium and Germany.
"Jarvis" for "iron people" -diabetics
JARVIS (Just Another Very Intelligent System) is the voice platform interface for diabetics Diabnext AI . Named after the butler of Iron Man, Tony Stark, it complements the CLIPSULIN system developed by the company. The system includes “smart” trackers that are mounted on an insulin pen and on a glucometer (a device for measuring blood sugar), a mobile application and a computer program for storing and analyzing records. In 2017, CLIPSULIN won the 2017 CES Innovation Award in the biotechnology category.
The system helps a diabetic to keep a daily diary of his condition: record insulin doses, meals eaten, and other parameters. A diary is needed for diabetics to control their blood sugar. Typically, entries are made into it manually - in special notebooks or in mobile applications. After each meal, the diabetic records the approximate amount of carbohydrates contained in it and the dose of insulin set for them. CLIPSULIN trackers allow you to do this automatically.
The principle of operation of the device is described in the video presented in the company's blog : a diabetic takes pictures of breakfast, and the image recognition system built into the tracker determines how much carbohydrate it contains. After that, the patient sets a dose on his insulin pen: the number of units is immediately fixed.
Another tracker is connected to the meter: using it, you can transfer data about each measurement of sugar to a mobile application and to a computer. You can also enter other parameters into the system: the results of standard tests, the level of glycated hemoglobin, data on weight, exercises, and so on.
At the same time, Jarvis can answer user questions and, like any neural network, it learns in the process of data collection. As the developers explain , if a person, for example, periodically eats a lot of pizza, AI will analyze how this type of food affects blood glucose levels and helps to calculate the dose of insulin more accurately. A doctor who has a persistent diabetic can also access Diabnext if the patient so desires.