Background technology
In busy daily life, people have no time to attend to the health status of oneself, until sick time how bad just find that the health of oneself has.When an ill, he both do not known sick which kind of, also do not know how this tackles the present situation, what uniquely can do is exactly go to see a doctor in time.But for some disease, before hospital, just process must tackled timely, such as heart disease and stroke.In addition, they do not know that the treatment of this disease is good at by which hospital yet, and which is bad at; Expense outside which hospital's medical insurance they can bear, which can not.Can not grasp enough there is the medical treatment resource information of reference value in, what they can do is only the hospital admission purposelessly arriving periphery, and this loses time in fact very much.
With regard to current, network has various medical resource website, but dragons and fishes jumbled together.Common medical resource recommended website is only carry out a basic diagnosis to the oral account of the state of an illness to disease according to the network user, and the reliability of this diagnosis is not high and depend on the state of an illness that user submits to a great extent and describe.And this kind of website is all the website of consulting class, accuracy and the reliability of its diagnosis are greatly limited to the doctor registered this website; In addition owing to being manual operation, promptness often also can not be guaranteed.
At present, wireless network, the fast development in the integrated chip field of physical sensors, embedded microcontroller and wave point allows real-time tracking health become possibility.These technology are all integrated in a Telemedicine System, can find early and process the abnormality of the healthy appearance of people, avoiding causing serious consequence.Many patients can benefit from long-term Monitoring Data, and these data can be diagnosis, and chronic disease situation and acute disease or operating recovery provide huge help.
Present Human Physiology sensor has had comparatively quantum jump, and for some physiological feature, as heart rate, body temperature, blood pressure etc. can have measured value accurately, and can upload in the middle of designated equipment in real time.The basic structure of next individual body biosensor point of normal conditions comprises processing module, sensor assembly and radio receiving transmitting module and power supply module.Its core is sensor assembly, for part physiology signal, because the complexity of body system and some physiological signals are weak signal, therefore for these physiological signals, draw one accurately and reliably measured value be comparatively difficulty.For a medical commending system, require that the human body signal parameter obtained is reliable and effective, therefore need when selection biosensor stability and the reliability of considering the data that this sensor obtains, to avoid because the unstability of hardware device and unreliability cause system finally to produce a larger error.
In addition, the present invention also according to the information gathered, uses the method and technology based on QoE to carry out medical resource recommendation targetedly.QoE is a kind of method that user of assessment is subject in the experience sense used after certain service, QoE lays particular emphasis on the experience sense of user to whole service and is subject to, the affecting parameters of QoE is the various factors that may occur in whole service process, and whether this service meets the expection of user.Introduce QoE, people can be assessed reliably to the expected value of a certain medical resource, satisfaction and service experience, ensure the accuracy that medical resource is recommended and quality with this.
Invention technology
The medical resource commending system that the present invention proposes instructs professional person to make accurate judgment the knowledge hierarchy of (how treating this disease) based on being similar to.The present invention uses medical treatment transducer to monitor the health status of people, and carries out real-time calculation and analysis, instructs and feedback for user provides according to analysis result.The present invention can also give a warning to user according to the condition etc. of the state of user, activity and living environment.In addition, the information of all records by the Internet transmission in medical server and the electronic health record being seamlessly integrated into user and research data base, can be inquired about when facilitating doctor necessary.
The present invention realizes by the following technical solutions.
A kind of health status real time monitor and diagnosis and treatment commending system, by human body biological characteristics sensor group, wireless network, personal server, medical server, medical resource supplier composition, the structure of Human Physiology sensor group is made up of processing module, sensor assembly and radio receiving transmitting module and power supply module;
By using the Partial Physiological Indexes of mankind's biometric sensor group to human body to monitor in real time, realizing data interaction by communicating between wireless network with personal server, forwarding the data to high in the clouds medical server by personal server again;
Personal server's process and integrated come the data of sensor group, carry out user health tracking, judge the health status of user, and provide Disease Warning Mechanism and guidance and medical resource recommendation according to health status; Personal server can perform some pretreatment and synchronous task, and communicates with long-range top service device; Personal server allows user by the suggestion of network reception from health care provider;
Health care provider is automatically collected the various data from user and is stored, by data integration in user's medical records; When user goes to hospital or be in other necessity, with reference to the history health status of user, in conjunction with the information that this sufferer the is up-to-date patient information known and relevant with other for sufferer provides appropriate guidance;
The various health monitoring data of server energy stored user, and by search have similar healthy time and unhealthy time the people of condition Monitoring Data, within the specific limits the likelihood ratio is carried out comparatively to the health degree of user according to Monitoring Data, use record and corresponding user experience quality to carry out medical resource recommendation by the medical resource of analogical object, and be presented on personal server.
Personal server of the present invention is a PC, a smart mobile phone or the equipment of custom-made.
Personal server of the present invention process and integrated come the data of sensor group, carry out user health tracking, judge the health status of user, and provide the method for Disease Warning Mechanism and guidance and medical resource recommendation to be according to health status:
Definition health state of user vector is F={f
1, f
2, f
3f
n, f
hibe i-th feature of health vector, N is the sum of feature; Consider that the impact of different feature f has significant difference, these features are quantified as different weights:
weight is larger then larger to health effect;
After each sensor gets a health vector f, health vector f is calculated as a health index according to the physiological feature weight sets preset by Human Physiology sensor group, health vector and health index is sent in the lump in cloud database afterwards by personal server; Data base, by the health detection data of a storage as much as possible people, realizes following the trail of the health of user;
Based on the detection data to user's individual every day, the health index of this user can be calculated:
The health index threshold value Th of a people is calculated from the data in certain limit
h; If health index I
hlower than definite threshold Th
h, then illustrate that this user has been in sub-health state, system will periodically be warned or remind this user and be provided some Improving advice, if health index is lower than Th
s, then illustrate that this user is sick, system then opens medical resource recommended program.
Searching of the present invention similar healthy time with the method for the people of condition Monitoring Data time unhealthy: the process finding the people similar to active user can be converted into the process that searching has similar health vector, and this step comprises two sub-steps:
Sub-step 1: similar under health status normal condition
People similar is under normal operation found based on health status is more similar under the healthy living condition of people;
This process can be described to:
For everyone, known health vector F
nif the health vector of targeted customer is F
th, calculate F
nand F
tnsimilarity;
Sim(F
n,F
tn)=1-Dis(F
n,F
tn)
Dis(F
n,F
tn)=||aF
n-aF
tn||
Dis (F
n, F
tn) be vectorial F
nand F
tndistance under normal operation;
Similar people can be selected by threshold value, then be chosen for similar users when similarity is greater than threshold value; After this similar users obtained is expressed as F
ns={ f
ns1, f
ns2;
Sub-step 2: similar under health status non-normal condition
From F under health status non-normal condition
nsin choose similar users, because similar physical condition information can be offered help for targeted customer solves health problem; This process can be described to:
For F
nsin everyone, known health vector F
aif the health vector of targeted customer is F
ta, according to following formulae discovery F
af
tasimilarity.
Sim(F
a,F
ta)=1-DiS(F
a,F
ta)
DiS(F
a,F
ta)=||aF
a-aF
ta||
D
is(F
a, F
ta) be vectorial F
aand F
tadistance under non-normal condition.
Similar people can be selected by threshold value, then be chosen for similar users when similarity is greater than threshold value; The similar users obtained is expressed as F
as={ f
as1, f
as2....
Medical resource by analogical object of the present invention uses record and corresponding user experience quality to carry out medical resource recommendation, and its method is:
If medical resource list is H
r={ h
r1, h
r2..., weighed by the QoE of medical resource, therefrom find out the medical resource of the most applicable targeted customer; For each medical resource h
r1, first quantize the QoE of user before using this resource, obtain QoE value list Q
b={ q
b1, q
b2..., and then the QoE after quantizing to use, obtain QoE value list Q
a={ q
a1, q
a2..., finally calculate the change numerical value of QoE; Suppose the QoE after the use of user be greater than use before QoE, and treatment time is shorter, so recommends this medical resource to user; Therefore calculated recommendation degree V
rand sort;
ΔQoE=|Q
a-Q
b|
Δ QoE is that user is using the QoE changing value before and after medical resource, and Δ t is the time that treatment continues;
Obtain the recommendation degree list V of medical resource
r={ v
r1, v
r2..., then to V
rsequence, is then pushed to user side according to this list by the medical resource corresponding to it and allows user oneself select most suitable resource.
The computational methods of QoE of the present invention for: the data that set Human Physiology sensor group can be collected calculate QoE, during monitoring, need the quantity continuing detection endorphins; Need the physiological signal of monitoring to comprise blood pressure, heart beating, brain wave can the physiological signal of representative of consumer response strength; Assuming that known current average endorphins quantity be e and current average physiological signal vector bp, h, bw}, and by with blood pressure, heart beating, the initial value of brain wave compares and calculates happy degree
E
i, bp
i, h
i, and bw
irepresent endorphins quantity, blood pressure, the initial value of heart beating and brain wave;
QoE score calculation formula is as follows:
qoe=ev×iv。
The present invention is divided into two parts, and Part I is physiology status monitoring, according to general knowledge, if no matter people are when normal and abnormal condition have closely similar physiological status, the probability that they have similar disease is very high, and therefore, the change of physiological status has very important reference value.In this part, we use mankind's biometric sensor, and the technology such as wireless network are monitored in real time to the Partial Physiological Indexes of human body, and real-time monitoring data is kept at high in the clouds, for the anticipation of disease provides Data support.In the present invention, required sensor is installed on human body by we, realizes data interaction, forward the data to high in the clouds again by personal server by the communication between sensor and personal server.Wherein personal server can make a PC, also can be a smart mobile phone or the equipment of custom-made.Part II is that medical resource is recommended, and we record the change of the QoE of all users after employing medical resource, then calculates meansigma methods as the threshold value of whether offering suggestions, then sort, recommend reference data to originate as medical resource.When an ill, we by search have similar healthy time and unhealthy time the people of condition Monitoring Data, so according to Monitoring Data, we can carry out the likelihood ratio comparatively to the health degree of user within the specific limits, use record and corresponding QoE to carry out medical resource recommendation by the medical resource of analogical object.
When user is sick, system provides health status relevant information and selectable instruction to user.If the no good cake of health status, this system provides the health care and medical resource list that are produced by the sequence of QoE value, and user can select the medical resource (as hospital and doctor) of corresponding specialty according to oneself current economic situation.
When user to Office visits or on the cycle of recovery time, medical resource commending system can assisted user rehabilitation, integrated system can obtain abundant information from multisensor, be warning user under user is in emergency, be in supervision recovery process or conventional activity user and feed back accordingly to user.A broad sense based on intelligence sensor be used for human detection medical resource commending system as shown in Figure 1.Intelligence sensor can be attached to skin surface or be placed in clothes or shoe inside is constantly collected and processes raw information, and they are sent to server.
For individual service, by medical biosensor human body health status, process and integrated (collection) carry out the data of sensor better to judge the health status of user, and it can provide disease early warning and guidance (such as between convalescence).Personal server can perform some pretreatment and synchronous task, and by home computer and long-range top service system communication.If user suffers from common disease (as flu), some nonprescription drugss can be provided to recommend.If be in a bad way, the healthcare resources list based on the sequence of QoE value can be provided to select professional medical resource (as hospital and doctor) for user according to the real economy ability of oneself, and personal server allows user by the suggestion of network reception from their health care provider.
For health care provider, health care provider automatically can be collected the various data from user and store, by data integration in user's medical records.When user goes to hospital or be in other necessity, can with reference to the history health status of user, in conjunction with the information that this sufferer the is up-to-date patient information known and relevant with other for sufferer provides appropriate guidance, and provide specialty based on QoE data, economic medical resource.
At present, medical monitoring is only applied to Countries, and just collects the healthy information of user passively, and can not provide corresponding medical advice based on user health information.Therefore the present invention has following advantage:
(1) for user provides the medical resource recommendation list based on QoE.This function is based on QoE and user can be helped to select most suitable professional medical mechanism.Patient can avoid blindly seeking medical advice.
(2) health care provider can get the health status record of patient, and these records can help doctor to make patient diagnoses more accurately.
Detailed description of the invention
Overall framework
The present invention is by health states (the such as post-operative recovery in the health status of Real-Time Monitoring people daily life or disease treatment, chronic disease), can find, with targeted customer, there is similar state in general calculation entironment, then keep healthy and medical resource list for it provides according to QoE value.If user is sick, he can obtain rapidly the most effectively treating according to the selection that user located in the past.As shown in Figure 2.
Health status is followed the trail of
Along with the develop rapidly of human biological's feature sensor, highly integrated low-power consumption, intelligent, the sensor node of miniaturization can be placed on inside and outside human body for monitoring human function and surrounding.Such as wireless body area network (WBAN) and HRS-I.The set of people biometric sensor with it and relevant sensor thereof is called an intelligent node by us, and intelligent node can measure such as shell temperature, the human body such as body kinematics and electrocardiosignal relevant information.Then this information is transmitted to the server in high in the clouds, this server stores a large amount of user characteristic data, diagnoses and output prescription for assist physicians.An intelligent node can provide about healthy abundant information, if a people has installed these sensors with it, the information that system can be collected according to sensor sets up his blood pressure image under normal circumstances in one day, even can set up the physiological feature situation of change in certain long period section.These information have very high reference value concerning doctor, help to reduce misdiagnosis rate.。It should be noted that, along with the change of health degree, the intensity of variation of different physiological features is inconsistent, therefore when the health evaluating to a people, different physiological features should be seized of different proportion, otherwise to assessment result then have larger deviation.We provide the mathematical expression of the method below.
The individual health information of monitoring is needed to comprise in this method: shell temperature, heart rate, blood pressure, cholesterol etc.Based on these information, we define health status vector is F={f
1, f
2, f
3f
n, f
hibe i-th feature of health vector, N is the sum of feature.Consider that the impact of different feature f has significant difference, these features are quantified as different weights:
weight is larger then larger to health effect.
Each can only after node gets a health vector f, health vector f can be calculated as a health index according to the physiological feature weight sets preset by intelligent node, health vector and health index is sent in the lump in cloud database afterwards.Data base is by the health detection data through a storage people that may be many, if because have a understanding when the analysis on the health status to a people to the physiological change situation of this people in a long period section, so this user health status the information revealing out also more horn of plenty, also will be more accurate to the diagnosis of the higher disease of some concealment.
Subhealth state is reported to the police
People is discrepant from birth, and environment and living habit all affect the physiological feature of a people, and so everyone healthy threshold value is different, and therefore for different people, we adopt different threshold values whether to judge its health.Threshold value obtains from a large amount of Monitoring Data learnings.Choosing of Monitoring Data should meet following 2 points, the first, and user is arranged in normal range in each physical signs current, or between certain stable region; The second, choosing of data should be enough, thus avoid and that bring error less due to data volume.
Based on the detection data to user's individual every day, we can calculate the health index of this user:
The health index threshold value Th of a people can be calculated from the data in certain limit
h.If health index I
hlower than definite threshold Th
h, then illustrate that this user has been in sub-health state, system will periodically be warned or remind this user and be provided some Improving advice.If health index is lower than Th
s, then illustrate that this user is sick, system then opens medical resource recommended program.
Search the people with similar health status
The object of our method is as possible for user provides the useful information relevant to their health status, can find the treatment mechanism and expert of suiting the medicine to the illness in the case of necessary accurately, avoid planless expending time in and energy.Carrying out before medical treatment recommends, we must find the people once with similar health status, then according to this person select QoE value in used medical resource higher user's related medical information recommend active user.First need the problem solved how to find a people once with similar health status.
The process finding the people similar to active user can be converted into the process found and have similar health vector.According to convention, the health status of a people can be divided into two parts, and one is healthy, and one is non-health.When the health status of a people is from then illustrating when normally becoming improper that this life is out of shape, consider from another angle, when two different people normally with improper two health status under time they all there is similar health vector time then they occur that the probability of identical health problem is very high.Recommend targetedly to carry out, we find M immediate health status before the people under general calculation entironment.
This step comprises two sub-steps:
Sub-step 1: similar under health status normal condition
In this step, we are more similar based on health status under the healthy living condition of people, and the change of their physiological status finds people similar under normal operation with regard to the general knowledge having more reference value.Such as, concerning more valuable than Europe of the health status information in other Asia Chinese.This process can be described to:
In this step, for everyone, we are known health vector F
nif the health vector of targeted customer is F
tn, calculate F
nand F
tnsimilarity.
Sim(F
n,F
tn)=1-Dis(F
n,F
tn)
Dis(F
n,F
tn)=||aF
n-aF
tn||
Dis (F
n, F
tn) be vectorial F
nand F
tndistance under normal operation.
Similar people can be selected by threshold value, then be chosen for similar users when similarity is greater than threshold value.After this similar users obtained is expressed as F by us
ns={ f
ns1, f
ns2.
Sub-step 2: similar under health status non-normal condition
This step in, we under health status non-normal condition from F
nsin choose similar users, because similar physical condition information can be offered help for targeted customer solves health problem.This process can be described to:
In this step, for F
nsin everyone, known health vector F
aif the health vector of targeted customer is F
ta, according to following formulae discovery F
af
tasimilarity.
Sim(F
a,F
ta)=1-Dis(F
a,F
ta)
Dis(F
n,F
tn)=||aF
a-aF
ta||
Dis (F
a, F
ta) be vectorial F
aand F
tadistance under non-normal condition.
Similar people can be selected by threshold value, then be chosen for similar users when similarity is greater than threshold value.After this similar users obtained is expressed as F by us
as={ f
as1, f
as2....
Recommend based on QoE medical resource
We obtain analogical object under non-normal condition and his diagnostic result, and therefore we can the type of preliminary forecasting disease, and system can estimate the risk of the disease of each type, then provides corresponding healthcare resources to recommend.
If health care resource list is H
r={ h
r1, h
r2..., such as hospital's name, we need the health care resource therefrom finding out the most applicable targeted customer.Can be weighed by the QoE of health care resource.For each health care resource h
r1, first quantize the QoE of people before using this resource, obtain QoE value list Q
b={ q
b1, q
b2..., then quantize QoE after usage, obtain QoE value list Q
a={ q
a1, q
a2..., finally calculate the change numerical value of QoE.QoE calculates please see next part.We QoE of hypothesis after the use of user is greater than the QoE before use, and treatment time is shorter, so recommends this medical resource to user.Therefore calculated recommendation degree V
rand sort.
ΔQoE=|Q
a-Q
b|
Δ QoE is that user is using the QoE changing value before and after medical resource, and Δ t is the time that treatment continues.
Obtain and keep healthy and the recommendation degree list V of medical resource
r={ v
r1, v
r2..., then to V
rsequence, is then pushed to user side according to this list by the medical resource corresponding to it and allows user oneself select most suitable resource.
QoE calculates
The numerical value of QoE determines the expection whether this medical resource meets this user, and from the side on show the health status whether this medical resource be suitable for this user.The measuring method of QoE has a variety of, and in the present system, we gather the data that Human Physiology sensor can collect and calculate QoE.
During monitoring, our method needs the quantity continuing to detect endorphins, and endorphins can represent level of happiness, because endorphins in excitement, can produce happiness and some physiological signals, the response strength of its representative of consumer in the processes such as love.
Need the physiological signal of monitoring to comprise blood pressure, heart beating, brain wave etc. can the physiological signal of representative of consumer response strength.Assuming that our known current average endorphins quantity is e and current average physiological signal vector { bp, h, bw}, all information all should meet normal form (between 0 to 1) when software application, by comparing the calculation of initial value happiness degree using before software that
E
i, bp
i, h
i, and bw
irepresent endorphins quantity, blood pressure, the initial value of heart beating and brain wave.
QoE score calculation formula is as follows:
qoe=ev×iv
Note that the QoE of calculating can also according to facial expression, speech, body language, voice messaging and physiological signal, and the situation of change using the health index after medical resource, all these methods can become a part for the framework of our commending system.