November 14th, today is the 28th United Nations Diabetes Day (World Diabetes Day). This year’s theme is - Prevent and control diabetes and protect your family. After all, the family is the basic unit of society. In family life, diabetic patients have similar eating habits to their family members. This means that diabetic patients must control their diet, but other members must also be prepared to prevent diabetes.
also shows that diabetes is no longer a patent for the elderly. In the past two years, diabetic patients have become younger. What’s more, the media previously reported that the youngest diabetic patient admitted to the hospital was only 7 years old.
Image source: pexels
In today's society, the lifestyle is becoming more and more convenient and the pace is getting faster and faster. Everyone is "fighting" to be a hundred-step youth, ordering takeout every day, irregular life schedule and other reasons are increasing the number of young people suffering from the disease. Diabetes Risk.
There are more and more young people with diabetes
According to the 8th edition of the Global Diabetes Map released by the International Diabetes Federation (IDF), approximately 425 million adults worldwide suffered from diabetes in 2017, of which 114 million were affected in China. The number of children and adolescents with type 1 diabetes (<20>
Image source: Pexels
On the other hand, judging from the growth rate, diabetes is growing faster among young people. A regional diabetes survey in 1996 showed that the prevalence of diabetes among people aged 20 to 40 was 2%, but a 2008 survey showed that this figure had reached 6%, an increase of three to four times, while the prevalence among the elderly has increased. The speed is only about 2 times.
It can be seen that the trend of younger patients with diabetes is obvious.
Diabetes itself is not terrible, but the complications
In fact, the most terrible thing about diabetes is not itself, but the various complications it can induce. According to statistics from the World Health Organization, there are more than 100 complications of diabetes, making it the disease with the most known complications. Diabetic eye disease is one of the most common complications. Such as macular degeneration, cataracts, retinal detachment and other eye diseases, among which retinopathy has become a major cause of acquired blindness in adults.
Because once this disease occurs, the patient's blinding progression is often irreversible. The prevalence of this complication in people with diabetes ranges from 21% to 36%. Compared with non- proliferative retinopathy , this proliferative retinopathy is more harmful to vision. The incidence rate increases with the increase of the disease time, and can lead to severe vision loss or even complete blindness.
Image source: pexels
According to the US National Diabetes Council report, the blindness rate of diabetic patients is 25 times higher than that of non-diabetic patients. However, the early treatment effect of diabetic retinopathy is still relatively ideal, and the cost of early prevention is much lower than the cost of late treatment. Therefore, diabetic patients should pay attention to the prevention of eye diseases.
On the other hand, there is a serious lack of resources for ophthalmologists. According to the China Health and Family Planning Statistical Yearbook, there were 537 ophthalmology hospitals nationwide in 2016, of which there were less than 40,000 ophthalmologists. These 40,000 people are not only responsible for treating diabetes. Patients with eye diseases also need to treat patients with other eye diseases, which shows that there is a serious asymmetry between patients and doctors.
AI-assisted diagnosis has benefited many patients with eye diseases
In this case, the advantages of artificial intelligence in the medical field are reflected. The machine can perform deep learning through fundus imaging data, forming image judgment and recognition capabilities similar to those of ophthalmologists, and assisting doctors in diagnosing eye diseases.
In order to ensure the accuracy of the final results, a huge amount of clinical data is needed as the basis. The advantages of eye specialist hospitals in this aspect are very obvious. Based on this, Aier Eye Group and Jinhong Technology Co., Ltd. (MiiS) jointly Developed artificial intelligence eye recognition solution.
Image source:Jinhong Technology official website
AIER and MiiS utilize thousands of privacy-protected images from AIRI data files through the MiiSHorus* digital handheld fundus camera (a portable device used to capture high-resolution images of the interior of the eye) equipment.The flexible Horus camera can be used in a variety of settings, such as community clinics, ambulances and hospital wards. ), which is trained through a large number of images to identify and distinguish key indicators of diabetic retinopathy and age-related macular degeneration. The huge data of
cannot be supported by the ordinary processor . Therefore, by using Intel ® Xeon ® processor in the back end of the instrument, Caffe* optimized for Intel® architecture (provided by Berkeley Visual Learning The open source deep learning framework developed by the Center (BVLC). Caffe optimized for Intel® architecture has now become an official branch of BVLC), the Intel® math kernel function library for deep neural networks, and multiple optimized open source mathematics and imaging. library to test its performance.
Although it is an entry-level product using Intel® Xeon® Scalable processors, MiiS and AI have also achieved significant speed improvements. The results show that the speed is more than 100 times faster than standard Caffe on the previous generation Intel® Xeon® processor. Optimized Caffe on Intel® Xeon® Bronze processors also increases the speed of performing the inference required to differentiate between patient conditions by more than 7.4x.
Previously, manual screening for diabetic reticulopathy and age-related macular degeneration could only achieve an average accuracy of 70% and 80%. In an analysis of imaging data of 5,000 AIDS patients, according to MiiS and AIDS Er reports that detection accuracy using deep learning solutions jumped to 93%. The application scenarios of
, a handheld fundus ophthalmoscope integrated with deep learning functions, will also be broader, allowing some community hospitals at the grassroots level to perform fundus screening on patients even without ophthalmologists, and then transmit the screening results to the solution. The inference system performs rapid analysis, categorizes and labels problem areas, and makes recommendations on whether to refer the patient to an eye specialist.
Image source: pexels
The rapid development of technology can assist doctors in diagnosing diseases, but for us, prevention is more important. Everyone with bad living habits needs to stay vigilant when diabetes comes to your door, keep your mouth shut, and move your legs, not just The chances of diabetes and other diseases coming to your doorstep will also be greatly reduced. I'm off work on Thursday, maybe you can consider swimming for fitness?
