TWI871509B - A physiological information prediction and integration system, method, and computer-readable medium thereof - Google Patents
A physiological information prediction and integration system, method, and computer-readable medium thereofDownload PDFInfo
- Publication number
- TWI871509B TWI871509BTW111113055ATW111113055ATWI871509BTW I871509 BTWI871509 BTW I871509BTW 111113055 ATW111113055 ATW 111113055ATW 111113055 ATW111113055 ATW 111113055ATW I871509 BTWI871509 BTW I871509B
- Authority
- TW
- Taiwan
- Prior art keywords
- heart rate
- information
- prediction
- module
- physiological
- Prior art date
Links
Images
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
Description
Translated fromChinese本發明係關於一種生理資訊預測與整合技術，尤其指一種基於人工智慧之生理資訊預測與整合系統、方法及其電腦可讀媒介。The present invention relates to a physiological information prediction and integration technology, and more particularly to a physiological information prediction and integration system, method and computer-readable medium based on artificial intelligence.
睡眠對於人們來說佔據了一生很大的一部分，且在很多實驗數據中也顯示睡眠的好壞會影響人體免疫力、記憶力及賀爾蒙分泌。但現代人由於生活壓力及工作繁重，造成心血管疾病及睡眠問題的人口數也以相當可觀的速度持續增加。對此，為了解決這些問題，在睡眠時的生理量測及監控更顯得重要。Sleep occupies a large part of people's lives, and many experimental data also show that the quality of sleep will affect the body's immunity, memory and hormone secretion. However, due to the pressure of life and heavy work, the number of people suffering from cardiovascular diseases and sleep problems in modern times is also increasing at a considerable rate. In order to solve these problems, physiological measurement and monitoring during sleep are even more important.
然而，目前市面上的心率監控裝置，多半是當使用者生理異常發生時才會發出警報，然而沒有辦法預先判斷使用者未來可能地身體狀況，也沒有系統性整合使用者所量測到的生理資訊，故現有技術無法讓使用者能預防潛在的心率異常問題。However, most of the heart rate monitoring devices currently on the market will only sound an alarm when the user has a physiological abnormality. However, there is no way to prejudge the user's possible future physical condition, nor is there a systematic integration of the user's measured physiological information. Therefore, existing technology cannot allow users to prevent potential heart rate abnormalities.
因此，如何提供一種生理資訊預測與整合機制，進而有效地監控及預測使用者生理情況，並能預防潛在的生理異常，遂成為業界亟待解決的課題。Therefore, how to provide a physiological information prediction and integration mechanism to effectively monitor and predict the user's physiological condition and prevent potential physiological abnormalities has become an issue that the industry needs to solve urgently.
為解決前述習知的技術問題或提供相關之功效，本發明提供一種生理資訊預測與整合系統，係包括：一生理量測裝置，係量測一受測者的一具有心電圖之心率資訊；以及一生理資訊整合裝置，係通訊連接該生理量測裝置以接收該心率資訊，且該生理資訊整合裝置包括：一心率擷取模組，係判斷該心率資訊是否為有效；一心率預測模組，係於該心率擷取模組判斷該心率資訊為有效時，由該心率預測模組對該心率資訊中之該心電圖進行資料分割，以於將該心電圖分割成複數子心電圖後，將該複數子心電圖分別轉換成複數組向量，以由該心率預測模組利用該複數組向量進行心率異常預測而產生一心率異常機率值之預測結果；及一心率訓練模組，係於該心率預測模組產生該預測結果後，由該心率訓練模組依據該心率資訊對該心率預測模組進行提升心率異常預測之準確率的訓練。In order to solve the above-mentioned known technical problems or provide related effects, the present invention provides a physiological information prediction and integration system, which includes: a physiological measurement device, which measures a subject's heart rate information with an electrocardiogram; and a physiological information integration device, which is communicatively connected to the physiological measurement device to receive the heart rate information, and the physiological information integration device includes: a heart rate acquisition module, which determines whether the heart rate information is valid; a heart rate prediction module, which is generated by the heart rate acquisition module when the heart rate acquisition module determines that the heart rate information is valid. The heart rate prediction module performs data segmentation on the electrocardiogram in the heart rate information, and after segmenting the electrocardiogram into a plurality of sub-electrocardiograms, the plurality of sub-electrocardiograms are converted into a plurality of sets of vectors respectively, so that the heart rate prediction module uses the plurality of sets of vectors to perform abnormal heart rate prediction and generate a prediction result of abnormal heart rate probability value; and a heart rate training module, after the heart rate prediction module generates the prediction result, the heart rate training module trains the heart rate prediction module according to the heart rate information to improve the accuracy of abnormal heart rate prediction.
本發明復提供一種生理資訊預測與整合方法，係包括：由一生理量測裝置量測一受測者的一具有心電圖之心率資訊；由一生理資訊整合裝置接收該心率資訊，且判斷該心率資訊是否為有效；於判斷該心率資訊為有效時，由該生理資訊整合裝置中之心率預測模組對該心率資訊中之該心電圖進行資料分割，以將該心電圖分割成複數子心電圖；由該心率預測模組將該複數子心電圖分別轉換成複數組向量，且利用該複數子心電圖進行心率異常預測而產生係為一心率異常機率值之預測結果；以及於該心率預測模組產生該預測結果後，由該生理資訊整合裝置中之心率訓練模組依據該心率資訊對該心率預測模組進行提升心率異常預測之準確率的訓練。The present invention further provides a physiological information prediction and integration method, comprising: measuring a subject's heart rate information having an electrocardiogram by a physiological measurement device; receiving the heart rate information by a physiological information integration device and determining whether the heart rate information is valid; when determining that the heart rate information is valid, performing data segmentation on the electrocardiogram in the heart rate information by a heart rate prediction module in the physiological information integration device, so as to segment the electrocardiogram into A complex sub-electrocardiogram; the heart rate prediction module converts the complex sub-electrocardiogram into a complex set of vectors, and uses the complex sub-electrocardiogram to predict heart rate abnormality to generate a prediction result of a heart rate abnormality probability value; and after the heart rate prediction module generates the prediction result, the heart rate training module in the physiological information integration device trains the heart rate prediction module according to the heart rate information to improve the accuracy of heart rate abnormality prediction.
於一實施例中，該生理資訊整合裝置更包括一加密模組，係加密該心率資訊及該預測結果，以產生並傳送一第一加密資訊及一第一加密碼。In one embodiment, the physiological information integration device further includes an encryption module that encrypts the heart rate information and the prediction result to generate and transmit a first encrypted information and a first encryption code.
於一實施例中，更包括一雲端伺服器，係通訊連接該生理資訊整合裝置，以接收來自該生理資訊整合裝置之該第一加密資訊及該第一加密碼，且該雲端伺服器加密該第一加密資訊，以產生一第二加密資訊及一第二加密碼。In one embodiment, a cloud server is further included, which is communicatively connected to the physiological information integration device to receive the first encrypted information and the first encryption code from the physiological information integration device, and the cloud server encrypts the first encrypted information to generate a second encrypted information and a second encryption code.
於一實施例中，該雲端伺服器儲存該第一加密資訊及該第二加密資訊，且將該第一加密碼及該第二加密碼儲存於一區塊鏈中。In one embodiment, the cloud server stores the first encrypted information and the second encrypted information, and stores the first encryption code and the second encryption code in a blockchain.
於一實施例中，該心率訓練模組透過一基於深度學習之強化學習方法在該心率預測模組產生該預測結果後，由該心率訓練模組即時地依據該心率資訊且採用該深度殘差網路對該心率預測模組中之預測模型進行提升心率異常預測之準確率的訓練。In one embodiment, the heart rate training module uses a reinforcement learning method based on deep learning to train the prediction model in the heart rate prediction module to improve the accuracy of abnormal heart rate prediction in real time based on the heart rate information and using the deep residual network after the heart rate prediction module generates the prediction result.
於一實施例中，該加密模組週期性地或即時地依據該生理資訊整合裝置之系統檔案或該系統檔案之版本資料產生一第一系統版本碼，且將該第一系統版本碼傳送至該雲端伺服器，以令該雲端伺服器比對該第一系統版本碼與一第二系統版本碼是否相同，進而於該第一系統版本碼與該第二系統版本碼不同時，由該雲端伺服器提供一更新系統檔案，並令該生理資訊整合裝置下載該更新系統檔案，以使該生理資訊整合裝置依據該更新系統檔案對該心率預測模組進行更新。In one embodiment, the encryption module periodically or instantly generates a first system version code according to the system file of the physiological information integration device or the version data of the system file, and transmits the first system version code to the cloud server, so that the cloud server compares the first system version code with a second system version code to see if they are the same. If the first system version code is different from the second system version code, the cloud server provides an updated system file, and the physiological information integration device downloads the updated system file, so that the physiological information integration device updates the heart rate prediction module according to the updated system file.
於一實施例中，該第二系統版本碼係由該雲端伺服器依據該更新系統檔案或該更新系統檔案之更新版本資料所計算之，且該雲端伺服器將該第二系統版本碼儲存於該區塊鏈中。In one embodiment, the second system version code is calculated by the cloud server based on the updated system file or the updated version data of the updated system file, and the cloud server stores the second system version code in the blockchain.
本發明又提供一種電腦可讀媒介，應用於具有處理器及/或記憶體的電腦或計算裝置中，該電腦或該計算裝置透過處理器及/或記憶體執行一目標程式及電腦可讀媒介，並用於執行電腦可讀媒介時執行如上所述之生理資訊預測與整合方法。The present invention also provides a computer-readable medium, which is applied to a computer or computing device having a processor and/or a memory. The computer or computing device executes a target program and a computer-readable medium through the processor and/or the memory, and is used to execute the physiological information prediction and integration method described above when executing the computer-readable medium.
由上可知，本發明之生理資訊預測與整合系統、方法及其電腦可讀媒介，主要藉由生理資訊整合裝置依據生理量測裝置對受測者所量測之心率資訊(亦即生理資訊)利用深度殘差網路進行心率異常預測，藉此產生一心率異常機率值之預測結果，以依據心率異常機率值判斷受測者潛在的生理異常，故相較於現有技術僅能於生理發生急劇變化的情況下，才能判斷出受測者身體發出異樣，本發明透過深度殘差網路藉此達到人工智慧之心率異常預測，且可即時監控受測者的心率變化，並透過人工智慧訓練心率預測模組，以預測受測者在未來會有心率異常的機率，進而提早做出預防行為。As can be seen from the above, the physiological information prediction and integration system, method and computer-readable medium of the present invention mainly use the deep residual network to predict heart rate abnormality based on the heart rate information (i.e., physiological information) measured by the physiological measurement device of the subject, thereby generating a prediction result of the probability value of heart rate abnormality, so as to judge the potential physiological abnormality of the subject based on the probability value of heart rate abnormality. Compared with the existing technology, which can only judge that the subject's body is abnormal when there are rapid physiological changes, the present invention achieves artificial intelligence prediction of heart rate abnormality through deep residual network, and can monitor the subject's heart rate changes in real time, and train the heart rate prediction module through artificial intelligence to predict the probability of the subject having an abnormal heart rate in the future, so as to take preventive actions in advance.
1:生理資訊預測與整合系統1: Physiological information prediction and integration system
10:生理量測裝置10: Physiological measurement device
11:量測模組11: Measurement module
12:顯示模組12: Display module
20:生理資訊整合裝置20: Physiological information integration device
21:心率擷取模組21: Heart rate acquisition module
22:心率訓練模組22: Heart rate training module
23:心率預測模組23: Heart rate prediction module
24:通知模組24: Notification module
25:加密模組25: Encryption module
26:第一儲存模組26: First storage module
27:通訊模組27: Communication module
30:雲端伺服器30: Cloud Server
31:第二儲存模組31: Second storage module
32:保密模組32: Confidentiality module
33:區塊鏈33: Blockchain
34:下載模組34: Download module
S21至S28、S31至S35、S41至S45:步驟S21 to S28, S31 to S35, S41 to S45: Steps
圖1係為本發明之生理資訊預測與整合系統之架構示意圖。Figure 1 is a schematic diagram of the architecture of the physiological information prediction and integration system of the present invention.
圖2係為本發明之心率辨識方法流程示意圖。Figure 2 is a schematic diagram of the heart rate recognition method process of the present invention.
圖3係為本發明之上傳心率資訊之方法流程示意圖。Figure 3 is a schematic diagram of the method flow for uploading heart rate information of the present invention.
圖4係為本發明之更新心率預測模組之方法流程示意圖。Figure 4 is a schematic diagram of the method flow for updating the heart rate prediction module of the present invention.
圖5係為本發明之心電圖之示意圖。Figure 5 is a schematic diagram of the electrocardiogram of the present invention.
以下藉由特定的具體實施例說明本發明之實施方式，熟悉此技藝之人士可由本說明書所揭示之內容輕易地瞭解本發明之其他優點及功效。The following is a specific and concrete example to illustrate the implementation of the present invention. People familiar with this technology can easily understand other advantages and effects of the present invention from the content disclosed in this manual.
須知，本說明書所附圖式所繪示之結構、比例、大小等，均僅用以配合說明書所揭示之內容，以供熟悉此技藝之人士之瞭解與閱讀，並非用以限定本發明可實施之限定條件，故不具技術上之實質意義，任何結構之修飾、比例關係之改變或大小之調整，在不影響本發明所能產生之功效及所能達成之目的下，均應仍落在本發明所揭示之技術內容得能涵蓋之範圍內。同時，本說明書中所引用之如「一」、「第一」、「第二」、「上」及「下」等之用語，亦僅為便於敘述之明瞭，而非用以限定本發明可實施之範圍，其相對關係之改變或調整，在無實質變更技術內容下，當視為本發明可實施之範疇。It should be noted that the structures, proportions, sizes, etc. depicted in the drawings attached to this specification are only used to match the contents disclosed in the specification for understanding and reading by people familiar with this technology, and are not used to limit the restrictive conditions for the implementation of the present invention. Therefore, they have no substantial technical significance. Any modification of the structure, change of the proportion relationship or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the effects and purposes that can be achieved by the present invention. At the same time, the terms such as "one", "first", "second", "upper" and "lower" used in this specification are only used to facilitate the clarity of the description, and are not used to limit the scope of the implementation of the present invention. The changes or adjustments in their relative relationships shall be regarded as the scope of the implementation of the present invention without substantially changing the technical content.
圖1係為本發明之生理資訊預測與整合系統1之架構示意圖。如圖1所示，生理資訊預測與整合系統1係包括：一生理量測裝置10、一生理資訊整合裝置20及一雲端伺服器30，其中，生理量測裝置10包括一量測模組11及一顯示模組12；生理資訊整合裝置20包括一心率擷取模組21、一心率訓練模組22、一心率預測模組23、一通知模組24、一加密模組25、一第一儲存模組26及一通訊模組27；以及雲端伺服器30包括一第二儲存模組31、一保密模組32、一區塊鏈33及一下載模組34。FIG1 is a schematic diagram of the structure of the physiological information prediction and
具體而言，生理量測裝置10係包含但不限於智慧型手錶或其他穿戴式電子裝置，生理資訊整合裝置20及雲端伺服器30係可建立於相同(或不同)伺服器(如通用型伺服器、檔案型伺服器、儲存單元型伺服器等)及電腦等具有適當演算機制之電子設備中，其中，生理量測裝置10、生理資訊整合裝置20及雲端伺服器30中之各個模組均可為軟體、硬體或韌體；若為硬體，則可為具有資料處理與運算能力之處理單元、處理器、電腦或伺服器；若為軟體或韌體，則可包括處理單元、處理器、電腦或伺服器可執行之指令，且可安裝於同一硬體裝置或分布於不同的複數硬體裝置。此外，生理資訊整合裝置20亦包含但不限於個人電腦、平板電腦、筆記型電腦等具有資料處理與運算能力之裝置。Specifically, the
另一方面，生理量測裝置10係透過有線或無線通訊技術，如蜂巢式網路、行動網路(4G、5G或6G)、Wi-Fi或藍牙等方式通訊連接生理資訊整合裝置20，而生理資訊整合裝置20亦能透過如上述之有線或無線通訊技術通訊連接雲端伺服器30。On the other hand, the
圖2係為本發明之心率辨識方法流程示意圖，且一併參閱圖1說明之，其中，該方法流程包含下列步驟S21至步驟S28：FIG2 is a schematic diagram of the heart rate recognition method process of the present invention, and is also explained in conjunction with FIG1 , wherein the method process includes the following steps S21 to S28:
於步驟S21中，生理量測裝置10係配戴於一受測者身上，且生理量測裝置10之量測模組11對在睡眠過程中之受測者進行心率量測，以於每間隔一時間(如1秒或5秒等)產生一心率資訊，而生理量測裝置10之顯示模組12將心率資訊顯示於生理量測裝置10之顯示螢幕上，其中，心率資訊係包含但不限於心跳次數、心電圖等。In step S21, the
於步驟S22中，生理資訊整合裝置20係接收來自生理資訊整合裝置20之心率資訊。In step S22, the physiological
於步驟S23中，生理資訊整合裝置20之心率擷取模組21擷取心率資訊，且判斷是否為有效的心率資訊，其中，若為無效的心率資訊，則回到步驟S22，重新接收來自生理資訊整合裝置20之心率資訊，例如：心率資訊中之心電圖的時間長度不足一預設時長(如1秒或5秒等)、心率資訊中之心電圖的數值異常(如心電圖中之電壓值皆為0)或心率資訊中之心電圖的電壓值超過量測範圍，以使心率資訊為無效；反之，若為有效的心率資訊，則生理資訊整合裝置20之第一儲存模組26儲存心率資訊，並執行步驟S24。In step S23, the heart
於步驟S24中，生理資訊整合裝置20之心率預測模組23將心率資訊中之心電圖依據心率格式進行資料分割，以將心電圖分割為複數子心電圖。例如：心率格式係為每100毫秒(ms)時長，故心率預測模組23將1秒鐘時長之心電圖分割為每100毫秒(ms)時長之複數子心電圖。In step S24, the heart
於步驟S25中，心率預測模組23確認複數子心電圖是否符合正確的心率格式，其中，若複數子心電圖並非正確的心率格式，則回到步驟S22，重新接收來自生理資訊整合裝置20之心率資訊；反之，若複數子心電圖係為正確的心率格式，則執行步驟S26。In step S25, the heart
於步驟S26中，心率預測模組23利用深度殘差網路(Deep residual network)所建立之預測模型依據複數子心電圖進行心率異常預測以產生一心率異常機率值之預測結果。In step S26, the heart
在一實施例中，心率預測模組23將複數子心電圖分別轉換成複數組向量，且複數組向量中分別包含有每個時點(如每毫秒)相對應之電壓值(mV)，再由心率預測模組23利用深度殘差網路所建立之預測模型依據複數組向量進行心率異常預測以產生心率異常機率值之預測結果，其中，心率異常機率值的範圍介於0~1之間。是以，本發明透過深度殘差網路藉此達到人工智慧之心率異常預測。In one embodiment, the heart
舉例而言，如圖5所示，心電圖利用組座標(x,y)呈現，且x代表時間毫秒(ms)，y代表毫伏特(mV)，心率預測模組23依據心電圖之時間將0~2000ms切成每100ms時長的20組子心電圖後，分別轉換成20組向量(如v1,v2,v3,…v20)，例如向量v1可表示為{(x1,y1),(x2,y2),(x3,y3)…}。再者，利用已經訓練好的預測模型之機率公式(1)計算出心率異常機率值(即機率P)，機率公式(1)如下所示：For example, as shown in FIG5 , the electrocardiogram is presented using a set of coordinates (x, y), and x represents time in milliseconds (ms), and y represents millivolts (mV). The heart
機率P=a1v1+a2v2+a3v3+…a20v20 (1)Probability P = a1v1 + a2v2 + a3v3 + ... a20v20 (1)
其中，a1~a20代表預測模型內部經訓練後的參數。Among them, a1~a20 represent the trained parameters within the prediction model.
於步驟S27中，生理資訊整合裝置20之通知模組24係將預測結果顯示於一顯示螢幕上以提供預測結果給受測者或是管理人員。在一實施例中，通知模組24亦可判斷當預測結果之心率異常機率值大於一機率閾值(如0.7)時，直接通報救護單位或是緊急聯絡人受測者身命體徵異常，以使救護單位或是緊急聯絡人把握黃金救援時間對受測者進行搶救。例如：通知模組24可在如智慧型手錶之生理量測裝置10上發送警示的推播訊息。In step S27, the
於步驟S28中，生理資訊整合裝置20之心率訓練模組22依據心率資訊對心率預測模組23進行訓練，以提升心率預測模組23之心率異常預測的準確率。In step S28, the heart rate training module 22 of the physiological
在一實施例中，心率訓練模組22是利用深度殘差網路對預測模型進行訓練，透過多層的類神經網路架構，及大量的心率資訊當作訓練輸入，且對心率資訊中之心電圖之時間軸進行切割，藉此對預測模型進行提升心率異常預測之準確率的訓練。再者，心率訓練模組22更可依據生理量測裝置10每間隔一時間(如1秒或5秒等)所產生並傳送之心率資訊以及藉由生理量測裝置10判斷出受測者之真實情況(即判斷受測者是否真的發生心律異常)，以確認預測結果是否正確，再透過基於深度學習之強化學習方法(Reinforcement learning)即時地依據接收到的心率資訊對預測模型進行訓練，以使預測模型可動態地強化，並改善其預測準確率。In one embodiment, the heart rate training module 22 uses a deep residual network to train the prediction model, through a multi-layer neural network architecture and a large amount of heart rate information as training input, and cuts the time axis of the electrocardiogram in the heart rate information, thereby training the prediction model to improve the accuracy of abnormal heart rate prediction. Furthermore, the heart rate training module 22 can determine the actual condition of the subject (i.e., determine whether the subject actually has arrhythmia) based on the heart rate information generated and transmitted by the
舉例而言，下列表1係為複數筆心率異常機率值及其對應之真實情況。如表1所示，右邊欄位係為心率異常機率值，左邊欄位係為生理量測裝置10確認受測者之真實情況，亦即受測者是否有發生心率異常顫動，其中，Ture係為心率異常顫動，而False係為心率正常。詳言之，假設以0.7當作機率閾值，當心率異常機率值小於0.7即預測未來不會發生心率異常顫動，而當心率異常機率值大於或等於0.7即預測未來會發生心率異常顫動。是以，如表1所示，預測模型預測出第5筆之心率異常機率(P=0.3333333)小於0.7，即判斷未來不會發生心率異常顫動，然而之後於生理量測裝置10對受測者量測心律時，得到受測者之真實情況係為發生心律異常顫動(Ture)，因此預測模型之預測結果產生錯誤。For example, the following Table 1 shows a plurality of abnormal heart rate probability values and their corresponding real situations. As shown in Table 1, the right column is the abnormal heart rate probability value, and the left column is the real situation of the subject confirmed by the
表1：心率異常機率值及其對應之真實情況
此時，心率訓練模組22依據第5筆之心率異常機率值所對應之心率資訊利用強化學習方法訓練預測模型，以透過這樣的觀測回饋修正預測模組的參數(例如a1~a20)，校正出更準確預測模組。At this time, the heart rate training module 22 uses the reinforcement learning method to train the prediction model based on the heart rate information corresponding to the fifth heart rate abnormal probability value, so as to correct the parameters of the prediction module (such as a1~a20) through such observation feedback to calibrate a more accurate prediction module.
圖3係為本發明之上傳心率資訊之方法流程示意圖，且一併參閱圖1及圖2說明之，其中，該方法流程包含下列步驟S31至步驟S35：FIG3 is a schematic diagram of the method flow of uploading heart rate information of the present invention, and is also explained with reference to FIG1 and FIG2 , wherein the method flow includes the following steps S31 to S35:
於步驟S31中，於生理資訊整合裝置20接收到心率資訊時，或於心率預測模組23得到預測結果後，生理資訊整合裝置20之加密模組25對一心率資訊及一預測結果進行加密以產生一第一加密資訊及一第一加密碼。In step S31, when the physiological
在一實施例中，加密模組25係利用雜湊演算法、對稱或非對稱加密演算法對心率資訊及預測結果進行加密以產生第一加密資訊，而第一加密碼可為經由雜湊演算法依據心率資訊及預測結果所計算之雜湊碼，或是第一加密碼可為對應於對稱或非對稱加密演算法之金鑰，其中，雜湊演算法係包含但不限於SHA-2、SHA-3等。In one embodiment, the
於步驟S32中，生理資訊整合裝置20之通訊模組27傳送一通行碼至一雲端伺服器30，以使雲端伺服器30依據通行碼驗證受測者身份及其合法性，以於通行碼正確無誤時，通訊模組27將第一加密資訊及第一加密碼透過有線或無線通訊技術傳送至雲端伺服器30。In step S32, the
於步驟S33中，雲端伺服器30之第二儲存模組31儲存第一加密資訊。In step S33, the
於步驟S34中，雲端伺服器30之保密模組32對第一加密資訊進行加密以產生一第二加密資訊及一第二加密碼。In step S34, the
在一實施例中，保密模組32係利用雜湊演算法、對稱或非對稱加密演算法對第一加密資訊進行加密以產生第二加密資訊，而第二加密碼可為經由雜湊演算法依據第一加密資訊所計算之雜湊碼，或是第二加密碼可為對應於對稱或非對稱加密演算法之金鑰，其中，雜湊演算法係包含但不限於SHA-2、SHA-3等。In one embodiment, the
於步驟S35中，保密模組32將第一加密碼及第二加密碼儲存於區塊鏈33中，而第二儲存模組31儲存第二加密資訊。是以，透過區塊鏈33具有不可竄改及不可逆之特性，係確保存於區塊鏈33中之第一加密碼及第二加密碼的完整性及正確性。此外，利用區塊鏈技術更能使駭客無法讓鏈上的各個節點達成共識(亦即同時竄改各個節點上的資料)，則可避免駭客竄改資料的行為以及資料集中儲存的風險。In step S35, the
在一實施例中，保密模組32可利用儲存區塊鏈33中之第一加密碼及第二加密碼分別對第一加密資訊及第二加密資訊進行解密，進而提供心率資訊及預測結果給管理人員。In one embodiment, the
圖4係為本發明之更新心率預測模組之方法流程示意圖，且一併參閱圖1至圖3說明之，其中，該方法流程包含下列步驟S41至步驟S45：FIG4 is a schematic diagram of the method flow of updating the heart rate prediction module of the present invention, and is also explained with reference to FIG1 to FIG3 , wherein the method flow includes the following steps S41 to S45:
於步驟S41中，一生理資訊整合裝置20之加密模組25週期性地(如每天、每個禮拜)依據生理資訊整合裝置20之系統檔案或該系統檔案之版本資料產生一第一系統版本碼，或是每當生理資訊整合裝置20執行時，其加密模組25會即時地依據生理資訊整合裝置20之系統檔案或該系統檔案之版本資料產生一第一系統版本碼。在一實施例中，加密模組25利用雜湊演算法依據生理資訊整合裝置20之系統檔案或其版本資料以計算出如雜湊碼之第一系統版本碼。In step S41, an
於步驟S42中，生理資訊整合裝置20之通訊模組27傳送一通行碼至一雲端伺服器30，使雲端伺服器30依據通行碼驗證受測者身份及其合法性，以於通行碼正確無誤時，通訊模組27將第一系統版本碼透過有線或無線通訊技術傳送至雲端伺服器30。In step S42, the
於步驟S43中，雲端伺服器30之區塊鏈33儲存第一系統版本碼，且下載模組34比對第一系統版本碼與儲存於區塊鏈33中之第二系統版本碼是否為相同，其中，若第一系統版本碼與第二系統版本碼相同時，代表生理資訊整合裝置20之系統檔案為最新，並執行步驟S44；反之，若第一系統版本碼與第二系統版本碼不同時，代表需要更新生理資訊整合裝置20之系統檔案，並執行步驟S45。In step S43, the
在一實施例中，第二系統版本碼係由保密模組32依據管理人員所提供之生理資訊整合裝置20之一更新系統檔案或該更新系統檔案之更新版本資料利用雜湊演算法所計算出如雜湊碼之第二系統版本碼。In one embodiment, the second system version code is a second system version code such as a hash code calculated by the
於步驟S44中，雲端伺服器30之下載模組34不提供更新系統檔案至生理資訊整合裝置20。In step S44, the
於步驟S45中，生理資訊整合裝置20之下載模組34將更新系統檔案傳送至通訊模組27，以令生理資訊整合裝置20依據更新系統檔案對心率預測模組23進行更新，進而確保心率預測模組23保持最新的版本，且提升穩定性及心率異常預測之準確度。In step S45, the
此外，本發明還揭示一種電腦可讀媒介，係應用於具有處理器(例如，CPU、GPU等)及/或記憶體的計算裝置或電腦中，且儲存有指令，並可利用此計算裝置或電腦透過處理器及/或記憶體執行此電腦可讀媒介，以於執行此電腦可讀媒介時執行上述之方法及各步驟。In addition, the present invention also discloses a computer-readable medium, which is applied to a computing device or computer having a processor (e.g., CPU, GPU, etc.) and/or a memory, and stores instructions, and the computing device or computer can execute the computer-readable medium through the processor and/or memory to execute the above-mentioned method and each step when executing the computer-readable medium.
下列係為本發明之生理資訊預測與整合系統之實施例，且一併參閱圖1及圖2說明之，且相同處不再贅述。The following is an embodiment of the physiological information prediction and integration system of the present invention, and is described together with FIG. 1 and FIG. 2, and the same parts will not be repeated.
於本實施例中，一受測者配戴如智慧型手錶之生理量測裝置10，以令量測模組11每間隔一時間(如2秒等)量測受測者之心率，產生一包含心電圖(如圖5所示之心率資訊，並將心率資訊傳送至如電腦之生理資訊整合裝置20。當生理資訊整合裝置20接收到心率資訊後，其心率擷取模組21擷取且判斷心率資訊是否為有效。In this embodiment, a subject wears a
再者，於心率資訊為有效時，生理資訊整合裝置20之心率預測模組23將心率資訊中之心電圖進行資料分割，以依據心率格式(如每250毫秒)將心電圖分割為每250毫秒(ms)時長之複數子心電圖。是以，心率預測模組23將複數子心電圖分別轉換成複數組向量，再利用深度殘差網路對複數組向量進行心率異常預測以產生一心率異常機率值之預測結果。最後，生理資訊整合裝置20之通知模組24係將預測結果顯示於一顯示螢幕上以提供預測結果給受測者或是管理人員。Furthermore, when the heart rate information is valid, the heart
此外，於完成心率異常預測後，生理資訊整合裝置20之心率訓練模組22依據此次用於預測之心率資訊利用深度殘差網路進行心率預測模組23的訓練，以提升心率預測模組23預測的準確率。甚者，心率訓練模組22更透過基於深度學習之強化學習方法以於心率預測模組23每次產生預測結果後，立即依據該次的心率資訊進行心率預測模組23的訓練，以達到動態地強化心率預測模組23之效果。In addition, after the abnormal heart rate prediction is completed, the heart rate training module 22 of the physiological
綜上所述，本發明之生理資訊預測與整合系統、方法及其電腦可讀媒介，透過生理資訊整合裝置依據生理量測裝置對受測者所量測之心率資訊(亦即生理資訊)利用深度殘差網路進行心率異常預測，藉此產生一心率異常機率值之預測結果，以依據心率異常機率值判斷受測者潛在的生理異常，故相較於現有技術僅能於生理發生急劇變化的情況下，才能判斷出受測者身體發出異樣，本發明透過深度殘差網路藉此達到人工智慧之心率異常預測，且可即時監控受測者的心率變化，並透過人工智慧訓練心率預測模組，以預測受測者在未來會有心率異常的機率，進而提早做出預防行為。In summary, the physiological information prediction and integration system, method and computer-readable medium of the present invention predicts heart rate abnormality based on the heart rate information (i.e., physiological information) measured by the physiological measurement device of the subject using a deep residual network, thereby generating a prediction result of a heart rate abnormality probability value, so as to judge the potential physiological abnormality of the subject based on the heart rate abnormality probability value. , compared to the existing technology that can only judge that the subject's body is abnormal when there are rapid physiological changes, the present invention uses a deep residual network to achieve artificial intelligence prediction of heart rate abnormalities, and can monitor the subject's heart rate changes in real time, and train the heart rate prediction module through artificial intelligence to predict the probability of the subject having an abnormal heart rate in the future, so as to take preventive actions in advance.
本發明之生理資訊預測與整合系統、方法及其電腦可讀媒介，係具備下列優點或技術功效。The physiological information prediction and integration system, method and computer-readable medium of the present invention have the following advantages or technical effects.
一、本發明之心率訓練模組透過深度殘差網路(Deep residual network)對心率預測模組進行訓練，透過多層的類神經網路架構及大量的心率資訊當作訓練輸入，以提升心率預測模組的心率異常預測之準確率。1. The heart rate training module of the present invention trains the heart rate prediction module through a deep residual network, using a multi-layer neural network architecture and a large amount of heart rate information as training input to improve the accuracy of the heart rate prediction module in predicting abnormal heart rates.
二、本發明之心率訓練模組更可依據生理量測裝置每間隔一時間所產生並傳送之心率資訊，以透過基於深度學習之強化學習方法(Reinforcement learning)即時地對心率預測模組進行心率異常預測之訓練，以使心率預測模組可動態地強化其預測準確率。2. The heart rate training module of the present invention can also train the heart rate prediction module for abnormal heart rate prediction in real time based on the heart rate information generated and transmitted by the physiological measurement device at intervals through a reinforcement learning method based on deep learning, so that the heart rate prediction module can dynamically enhance its prediction accuracy.
三、本發明透過將經由雜湊演算法所計算出之第一加密碼、第二加密碼及第二系統版本碼，或是如對稱或非對稱加密演算法之金鑰之第一加密碼及第二加密碼紀錄於一區塊鏈中，透過區塊鏈具有不可竄改及不可逆之特性，可以確保存於區塊鏈中之第一加密碼、第二加密碼及第二系統版本碼的完整性及正確性。此外，利用區塊鏈技術更能使駭客無法讓鏈上的各個節點達成共識(亦即同時竄改各個節點上的資料)，則可避免駭客竄改資料的行為以及資料集中儲存的風險。3. The present invention records the first encryption code, the second encryption code and the second system version code calculated by the hashing algorithm, or the first encryption code and the second encryption code of the key of the symmetric or asymmetric encryption algorithm in a blockchain. The blockchain has the characteristics of being unalterable and irreversible, so the integrity and correctness of the first encryption code, the second encryption code and the second system version code stored in the blockchain can be ensured. In addition, the use of blockchain technology can prevent hackers from reaching a consensus on the nodes on the chain (that is, modifying the data on each node at the same time), thereby avoiding the behavior of hackers modifying data and the risk of centralized data storage.
四、本發明透過生理資訊整合裝置週期性地上傳其第一系統版本碼至雲端伺服器，進而確認生理資訊整合裝置之系統版本是否為最新，以於生理資訊整合裝置之系統版本並非為最新系統版本時，雲端伺服器提供一更新系統檔案以更新生理資訊整合裝置進行系統更新，進而提升心率預測模組之穩定性及心率異常預測之準確度。4. The present invention periodically uploads the first system version code of the physiological information integration device to the cloud server to confirm whether the system version of the physiological information integration device is the latest. When the system version of the physiological information integration device is not the latest system version, the cloud server provides an update system file to update the physiological information integration device for system update, thereby improving the stability of the heart rate prediction module and the accuracy of abnormal heart rate prediction.
上述實施形態僅例示性說明本發明之原理及其功效，而非用於限制本發明。任何熟習此項技藝之人士均可在不違背本發明之精神及範疇下，對上述實施形態進行修飾與改變。因此，本發明之權利保護範圍應如申請專利範圍所列。The above implementation forms are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with this art may modify and change the above implementation forms without violating the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be as listed in the scope of the patent application.
1:生理資訊預測與整合系統1: Physiological information prediction and integration system
10:生理量測裝置10: Physiological measurement device
11:量測模組11: Measurement module
12:顯示模組12: Display module
20:生理資訊整合裝置20: Physiological information integration device
21:心率擷取模組21: Heart rate acquisition module
22:心率訓練模組22: Heart rate training module
23:心率預測模組23: Heart rate prediction module
24:通知模組24: Notification module
25:加密模組25: Encryption module
26:第一儲存模組26: First storage module
27:通訊模組27: Communication module
30:雲端伺服器30: Cloud Server
31:第二儲存模組31: Second storage module
32:保密模組32: Confidentiality module
33:區塊鏈33: Blockchain
34:下載模組34: Download module
Claims (5)
Translated fromChinesePriority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW111113055ATWI871509B (en) | 2022-04-06 | 2022-04-06 | A physiological information prediction and integration system, method, and computer-readable medium thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW111113055ATWI871509B (en) | 2022-04-06 | 2022-04-06 | A physiological information prediction and integration system, method, and computer-readable medium thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TW202341168A TW202341168A (en) | 2023-10-16 |
| TWI871509Btrue TWI871509B (en) | 2025-02-01 |
Family
ID=89856074
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW111113055ATWI871509B (en) | 2022-04-06 | 2022-04-06 | A physiological information prediction and integration system, method, and computer-readable medium thereof |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI871509B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWM466300U (en)* | 2007-08-30 | 2013-11-21 | Fego Prec Ind Co Ltd | System for integrating and managing health related information |
| TW201417030A (en)* | 2012-10-18 | 2014-05-01 | Footwear & Recreation Technology Res Inst | Fitness machine information management system and method |
| CN107981858A (en)* | 2017-11-27 | 2018-05-04 | 乐普（北京）医疗器械股份有限公司 | Electrocardiogram heartbeat automatic recognition classification method based on artificial intelligence |
| US20180203978A1 (en)* | 2017-01-13 | 2018-07-19 | Microsoft Technology Licensing, Llc | Machine-learning models for predicting decompensation risk |
| US20190038148A1 (en)* | 2013-12-12 | 2019-02-07 | Alivecor, Inc. | Health with a mobile device |
| CN110570197A (en)* | 2019-09-17 | 2019-12-13 | 腾讯科技（深圳）有限公司 | Data processing method and device based on block chain |
| CN112104692A (en)* | 2020-06-29 | 2020-12-18 | 黑龙江省医院 | Medical Internet of things health monitoring method |
| US20210007659A1 (en)* | 2019-06-28 | 2021-01-14 | Teleplus Healthcare LLC | System and method for sleep disorders: screening, testing and management |
| CN112617855A (en)* | 2020-12-31 | 2021-04-09 | 平安科技（深圳）有限公司 | Electrocardiogram analysis method and device based on federal learning and related equipment |
| US20210127985A1 (en)* | 2019-11-02 | 2021-05-06 | West Affum Holdings Corp. | Secure Patient Data |
| TWM612245U (en)* | 2021-01-12 | 2021-05-21 | 臺北榮民總醫院 | Real-time monitoring system for continuous medical data streaming |
| US20210407678A1 (en)* | 2020-06-24 | 2021-12-30 | Neuropace, Inc. | Systems and methods for using federated learning for training centralized seizure detection and prediction models on decentralized datasets |
- 2022
- 2022-04-06TWTW111113055Apatent/TWI871509B/enactive
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWM466300U (en)* | 2007-08-30 | 2013-11-21 | Fego Prec Ind Co Ltd | System for integrating and managing health related information |
| TW201417030A (en)* | 2012-10-18 | 2014-05-01 | Footwear & Recreation Technology Res Inst | Fitness machine information management system and method |
| US20190038148A1 (en)* | 2013-12-12 | 2019-02-07 | Alivecor, Inc. | Health with a mobile device |
| US20180203978A1 (en)* | 2017-01-13 | 2018-07-19 | Microsoft Technology Licensing, Llc | Machine-learning models for predicting decompensation risk |
| CN107981858A (en)* | 2017-11-27 | 2018-05-04 | 乐普（北京）医疗器械股份有限公司 | Electrocardiogram heartbeat automatic recognition classification method based on artificial intelligence |
| US20210007659A1 (en)* | 2019-06-28 | 2021-01-14 | Teleplus Healthcare LLC | System and method for sleep disorders: screening, testing and management |
| CN110570197A (en)* | 2019-09-17 | 2019-12-13 | 腾讯科技（深圳）有限公司 | Data processing method and device based on block chain |
| US20210127985A1 (en)* | 2019-11-02 | 2021-05-06 | West Affum Holdings Corp. | Secure Patient Data |
| US20210407678A1 (en)* | 2020-06-24 | 2021-12-30 | Neuropace, Inc. | Systems and methods for using federated learning for training centralized seizure detection and prediction models on decentralized datasets |
| CN112104692A (en)* | 2020-06-29 | 2020-12-18 | 黑龙江省医院 | Medical Internet of things health monitoring method |
| CN112617855A (en)* | 2020-12-31 | 2021-04-09 | 平安科技（深圳）有限公司 | Electrocardiogram analysis method and device based on federal learning and related equipment |
| TWM612245U (en)* | 2021-01-12 | 2021-05-21 | 臺北榮民總醫院 | Real-time monitoring system for continuous medical data streaming |
Non-Patent Citations (2)
| Title |
|---|
| 期刊 Abdur Rahim Forkan and Ibrahim Khalil, "A probabilistic model for early prediction of abnormal clinical events using vital sign correlations in home-based monitoring" 2016 IEEE International Conference on Pervasive Computing and Communications, 2016,;* |
| 期刊 Dang Hao, etc., "A deep biometric recognition and diagnosis network with residual learning for arrhythmia screening using electrocardiogram recordings", IEEE Access, 2020 Aug 17, pp. 153436-54* |
Also Published As
| Publication number | Publication date |
|---|---|
| TW202341168A (en) | 2023-10-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190217090A1 (en) | Pulsed Electromagnetic Field Tissue Stimulation Treatment and Compliance Monitoring | |
| US10929510B2 (en) | Patient care systems employing control devices to identify and configure sensor devices for patients | |
| US11551793B2 (en) | Systems and methods for optimizing management of patients with medical devices and monitoring compliance | |
| CN115769302A (en) | Epidemic Surveillance System | |
| GB2436721A (en) | Automated method for adapting the settings of a patient monitor | |
| US20250316375A1 (en) | Glucose prediction using machine learning and time series glucose measurements | |
| JP2023536038A (en) | Group disease identification using wearable glucose monitoring devices | |
| KR20170057668A (en) | Electronic system and electronic device | |
| CN108053875A (en) | Nursing plan management method, device, medium and electronic equipment | |
| US20220095973A1 (en) | Systems, methods, and devices for monitoring stress associated with electronic device usage and providing interventions | |
| CN112635062A (en) | Data processing method and device based on block chain, electronic equipment and storage medium | |
| CN116848584A (en) | AI-enabled healthcare service access | |
| Ahamed et al. | Design of an energy-efficient IOT device-assisted wearable sensor platform for healthcare data management | |
| CA3166418C (en) | Compensation for missing readings from a glucose monitor in an automated insulin delivery system | |
| CN115148379A (en) | System and method for realizing intelligent health monitoring of solitary old people by utilizing edge calculation | |
| TWI871509B (en) | A physiological information prediction and integration system, method, and computer-readable medium thereof | |
| Yuvaraj et al. | Predicting Hand Injury Severity Using Bayesian Networks in Cloud-Based Emergency Medicine | |
| US11904179B2 (en) | Virtual reality headset and system for delivering an individualized therapy session | |
| CN111446001A (en) | Intelligent system and method for crowd dynamic medical observation and isolation management | |
| JP2022140935A (en) | Heat stroke prediction device, heat stroke prediction method, and program | |
| KR20220106619A (en) | Electronic device for performing federated learning using hardware security architecture and federated learning method using the thereof | |
| Agarwal et al. | IoT-Based ECG monitoring system for health care applications | |
| JP2021064242A (en) | System, method, information processing device and program | |
| US20230033093A1 (en) | Systems and methods for remote measurement of cervical range of motion | |
| CN109815443A (en) | A Statistical Evaluation System for Production Safety Accidents |