Recently, outpatient service guides examine robot " young doctor " hospital of Han Dan downtown is visited in Heibei east hall of area outpatient service is formal mount guard, slow-witted bud appearance and interesting and interactive attract many patients to surround view. Hao Qunying is photographed
"Medical skill " exceed young doctor
Meeting " look " video, meeting " read " anamnesis, meeting " perform an operation " , meeting " do " examination, still can give out clinical diagnostic proposal; "Medical skill " exceed young doctor, a few domains can compare a shoulder with senior doctor. It, namely " artificial intelligence doctor " .
As human doctor, "Artificial intelligence doctor " also be through looking, hear, listen wait for a method to see a doctor.
It is with lung tubercle exemple, small the focus of a disease to 1 millimeter, read a doctor to need one Zhang Zhang sees CT video picture will search, conclude a size, density. Senior read a cure the life all reads 1 piece 10 minutes, large hospital everyday the film exceeds 100 thousand pieces, the work that reviews a doctor is nervous and onerous. Nowadays, a few hospitals begin to introduce sieve of artificial intelligence system to check, read a time to fall to 1 minute.
"Artificial intelligence doctor " not only efficiency is very tall, in respect of diagnose a disease more careful, comprehensive. In Hua Zhongke ability university is the same as aid college of medicine accessary assist with the hospital, the artificial intelligence system that develops by the medical treatment that depend on a graph can detect not only lobar tubercle focus of a disease, still can undertake to character of focus of a disease multidimensional spend a description, include cost of size, bulk, density, CT, nodal token can cover 6 kinds of common fine evil sex character to resemble -- cent leaf, burr, pleura caves, empty, vacuole, calcify. The system of healthy development in A sifts circumjacent disease together check, include lobar path bubble, arteriosclerosis, lymphatic belt to change, lobar density heighten, all alone etc.
Artificial intelligence mount " eye " , can read the picture of standardization, sift fish focus of a disease. University of Hua Zhongke ability is the same as aid college of medicine accessary assist with the hospital clinical trial this " artificial intelligence doctor " , discover its check goes out rate amount to 95.78% , rate of false alarm is only however 2.63% . 2018, doctor of 60 video families passes this courtyard AI system is sentenced read video case of illness to exceed 150 thousand.
Besides, "Artificial intelligence doctor " still can check the disease such as tumour of rectum of esophagus cancer, diabetic retinal pathological changes, knot, breast cancer, still can investigate children character even age, technical level no less than senior doctors.
Artificial intelligence still has delicacy " auditive " . In in relief area of cottage of city of Hefei of the Anhui province, division big question flies wisdom cure assistant already served orgnaization mount guard in community sanitation 2018, in process of communication of doctors and patients, wisdom cure assistant adopts big data and intelligent speech technique, generate and collect medical record automatically, the doctor still can inquire likeness case of illness, clinical guideline and to disease medicines and chemical reagents. Current, this system already completed the auxiliary diagnostic proposal of 7000 more than person-time.
Recently, "Artificial intelligence doctor " return mount " cerebrum " . In Guangzhou Fu center, artificial intelligence system learns " understand " anamnesis, resemble human doctor next same, give visit a patient at home to break. Medical history of disease of action in chief of doctor general patient, symptom, individual, examination examines result, video learn to check the information such as the result, circumstance that use drug to input anamnesis text, the system changes free anamnesis text automatically into the data of standardization, standardization and structuralization. Artificial intelligence system " understand " after anamnesis, decide an outcome to visit a patient at home again.
"Artificial intelligence doctor " is diagnostic accuracy rate high? It is with respiratory system disease exemple, this artificial intelligence is mixed to disease of the upper respiratory tract the diagnostic accuracy rate of next respiratory tract diseases is respectively 89% with 87% , asthmatic to different type diagnostic accuracy rate is in 83% to 97% between.
Pass ceaseless training " artificial intelligence doctor " , "Eye " " auditive " " cerebrum " develop increasingly, involve disease to plant increasing, field is wider and wider, include clinical assistant, auxiliary diagnosis and treatment, medicine video, gene detects, health management.
Artificial intelligence relies on huge data
The standardization degree of each course data, affecting the applied rate of artificial intelligence. Each hospital equipment are different, data dimension is spent different also
The doctor grows skill, rely on study of medical major academy, 2 rely on clinical experience to accumulate. "Artificial intelligence doctor " what to rely on? Rely on ability of computation of huge data, cloud. "Eat " be over after data, the course trains clinical thinking ceaselessly, the system can resemble human doctor seeing a doctor euqally.
"Eat " after huge data, the machine can become a doctor not only, and can do scientific research, education, management to wait, help doctor and level of hospital promotion scientific research, increase capacity of diagnosis and treatment.
In Sichuan courtyard of university China Western medicine, the whole dimension that brings into this courtyard to received treated lung cancer patient up to now 2009 according to plan medical treatment spends desensitization clinical data, get through is clinical the many systems data such as outpatient service, be in hospital, anamnesis, pathology, built first lung cancer of domestic disease of intelligence of clinical scientific research is planted library. Had this disease to plant library, the capability of many section office research related to lung cancer diagnosis and treatment promote the hospital greatly, orgnaization of body of other doctor couplet also benefits a lot.
Medical treatment of healthy artificial intelligence upgrades to 2 version in A, besides clinical, still have function of platform of text scientific research, video scientific research, offer operation of imitate of fictitious patient, VR to be used at education.
Village village clinic holds the post of in Henan Jia county, the reporter saw small doctor artificial intelligence complementary examine system -- whole family assists system of diagnosis and treatment of intelligence of cure of the system of diagnosis and treatment, Taichun that hang crock. Village cure Zhang Qiaofen inputs basic symptom of the patient, medical history simply to wait, can see relevant danger weighs ill, common disease to may hint immediately. "We are very few at ordinary times danger of bring into contact with weighs disease, but still worry in the heart in case by accident examine, can delay villager treatment. Can delay villager treatment..
According to introducing, this one whole family assists system of diagnosis and treatment to surpass anamnesis of portion of 5 million document, ten million and healthy record through learning, already enclothed 2000 many disease to plant at present, 5000 many symptoms, hit the target rate achieve 90% . Accumulative total of system of diagnosis and treatment of intelligence of cure of the Taichun that hang crock assists evolution quantity to already exceeded 2 million pieces.
"Capability of service of basic level doctor is not strong, artificial intelligence assists diagnosis and treatment to be able to remedy the problem of resource scarcity, promotion doctor serves a level. " population of courtyard of Chinese company division and Chen Qiulin of director of lab of social security of labor economy institute think, medical treatment artificial intelligence can raise the essence that medical treatment diagnoses to follow rate, also can replace the operation in operation of a few high-grade, OK still partial manpower resource is replaced in service of a few medical treatment, reduce medical treatment cost thereby.
The data of artificial intelligence study from clinical come, return so that changeover becomes structured form, make a model next, according to training of thinking of clinical diagnosis and treatment, study, cipher out result. Data is crucial, the standardization degree of each course data, affecting the applied rate of artificial intelligence.
Ni Hao of president of the medical treatment that depend on a graph tells a reporter, medical treatment data not the problem that the standard is a catholicity. Although video it is a batch of data with better standardization, but different hospital or difference are very big. Each hospital equipment are different, data dimension is spent different also. The data of high quality is very scarce, need costs more algorithm, use data structuralization talent first.
2018, ning Guang of assistant dean of accessary Ruijin hospital guides college of medicine of university of traffic of academician of Chinese Academy of Engineering, Shanghai lab of healthy artificial intelligence is collective in group and A research and development " luck aids candy rather " artificial intelligence doctor. In advancing a process, ning Guang also discovered the problem of data, if criterion numeral occupies lack, disease diagnostic standard is skimble-scamble, along with visit data to scatter in each hospitals, data general comfortable sex is poorer wait.
Data standardizes degree and course maturity degree, diagnostic place to need exterior condition to concern. For instance video domain from start even if unite a standard, digitlization development rate is higher also; Dermatological department is diagnosed rely on picture and video to identify focus of a disease to wait a moment quite, development of these course artificial intelligence are rapidder.
Machine and doctor see a doctor in coordination
Medical treatment just is not diagnosed and treat, still involve doctor and patient between interactive, especially the doctor is right of the patient comfort have the effect that cannot replace
Artificial intelligence is medical, whether can you replace a doctor? OK and affirmative, return at present won't.
2017, artificial intelligence of new generation of the State Council plans to put forward, "Assistant of diagnosis and treatment of the operation robot that development man-machine cooperates with, intelligence " " man-machine of research and development cooperates with plan of clinical intelligence diagnosis and treatment " . This is meant, artificial intelligence is the doctor's assistant only.
Personnel of research and development of a few artificial intelligence puts forward, understand the doctor's psychology and clinical thinking only, let artificial intelligence learn this kind of thinking, just be true medical treatment artificial intelligence. However, this difficulty appears bad breakthrough.
"I undertake diagnostic to coming completely by the machine, support certain distrustful attitude, future still needs examine further. Just be not diagnosed because of medical treatment and treat, still involve doctor and patient between interactive, especially the doctor is right of the patient comfort have the effect that cannot replace. " Chen Qiulin says.
Future, "Artificial intelligence doctor " with human doctor perhaps a mount guard works. The reporter experienced mode of this kind of service. On platform of Beijing video cloud, 30 patients that artificial intelligence system uploads to basic level hospital are close CT of tubercle of 9000 pieces of lung is video undertake intelligence detects and identify, the doubt that sifts the first round fish is like nodal mark to come out, as auxiliary diagnostic result, offer doctor of 4 radiative families to undertake checkup. After the doctor is examined, think to be able to be adopted, sign to the report namely.
In this kind of new service pattern, still do natively by cure final and decision-making. A few clinicians express: Must ensure technology of artificial intelligence product is excellent above all, give out reasonable diagnostic proposal; Undertake grooming even next, change idea, get used to new service pattern. Of the doctor approbate and guide, will raise a patient to be opposite the degree of belief of artificial intelligence system.
Current, the development of industry of medical treatment artificial intelligence still faces a problem. "Each domain data do not have medical treatment interconnection each other is connected, what form finally is data only big, is not big data. Medical treatment artificial intelligence needs medical treatment talented person already, also need artificial intelligence talented person. Current, development compares a few projects with good first-rate perhaps business, be advanced by the talent of this two respects. " Chen Qiulin says, because involve the protection of individual privacy, be necessary to limit the property right of medical treatment data, property right clarity is helpful for realizing interconnection each other to connect.
Can foreknow, prospective mankind will not leave " artificial intelligence doctor " . The medical treatment in those days is no longer " medical treatment queueing up " , however " second medical treatment " " essence of life allows medical treatment " " individual character medical treatment " . (cartography of reporter Li Gongmei: Cai Huawei)
