[go: up one dir, main page]

TR2023009498A2 - BLOOD PRESSURE ESTIMATION SYSTEM - Google Patents

BLOOD PRESSURE ESTIMATION SYSTEM Download PDF

Info

Publication number
TR2023009498A2
TR2023009498A2 TR2023/009498A TR2023009498A TR2023009498A2 TR 2023009498 A2 TR2023009498 A2 TR 2023009498A2 TR 2023/009498 A TR2023/009498 A TR 2023/009498A TR 2023009498 A TR2023009498 A TR 2023009498A TR 2023009498 A2 TR2023009498 A2 TR 2023009498A2
Authority
TR
Turkey
Prior art keywords
blood pressure
ecg signals
machine learning
real
ecg
Prior art date
Application number
TR2023/009498A
Other languages
Turkish (tr)
Inventor
Kahraman Hakan
Original Assignee
Livewell Giyilebilir Saglik Ueruen Hizmet Ve Teknolojileri Sanayi Ve Ticaret Anonim Sirketi
Livewell Gi̇yi̇lebi̇li̇r Sağlik Ürün Hi̇zmet Ve Teknoloji̇leri̇ Sanayi̇ Ve Ti̇caret Anoni̇m Şi̇rketi̇
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Livewell Giyilebilir Saglik Ueruen Hizmet Ve Teknolojileri Sanayi Ve Ticaret Anonim Sirketi, Livewell Gi̇yi̇lebi̇li̇r Sağlik Ürün Hi̇zmet Ve Teknoloji̇leri̇ Sanayi̇ Ve Ti̇caret Anoni̇m Şi̇rketi̇ filed Critical Livewell Giyilebilir Saglik Ueruen Hizmet Ve Teknolojileri Sanayi Ve Ticaret Anonim Sirketi
Priority to TR2023/009498A priority Critical patent/TR2023009498A2/en
Publication of TR2023009498A2 publication Critical patent/TR2023009498A2/en
Priority to PCT/TR2024/050889 priority patent/WO2025034186A1/en

Links

Classifications

    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording for evaluating the cardiovascular system, e.g. pulse, heart rate, blood pressure or blood flow
    • A61B5/021Measuring pressure in heart or blood vessels
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/24Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316Modalities, i.e. specific diagnostic methods
    • A61B5/318Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/72Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235Details of waveform analysis
    • A61B5/7264Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • Cardiology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Artificial Intelligence (AREA)
  • Physiology (AREA)
  • Evolutionary Computation (AREA)
  • Signal Processing (AREA)
  • Psychiatry (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Fuzzy Systems (AREA)
  • Vascular Medicine (AREA)
  • Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)

Abstract

Buluş makine öğrenmesi ile EKG sinyallerinden invaziv olmayan kan basıncı tahmin sistemi (10) ile ilgilidir. Yenilik olarak kullanıcıdan anlık EKG sinyallerinin alınmasını sağlayan bir giriş birimini (14) ve alınan EKG sinyalinden tahmini kan basıncı değerini hesaplayan bir makine öğrenmesi modelini saklayan bir hafıza biriminden (11) aldığı sonuçları bir kullanıcı arayüzüne aktaran bir işlemci birimini (12) içermektedir ve de bahsedilen makine öğrenmesi modeli; hafıza birimine (11) girilmiş gerçek EKG sinyallerinden ve anlık EKG sinyallerinden karmaşıklık analizi ile sinyalin özniteliklerini çıkartacak, çıkartılan öznitelikleri istifleme regresyon modeli ile işleyecek ve anlık EKG sinyaline karşılık gelen tahmini kan basıncı değerini hesaplarken bahsedilen gerçek EKG sinyallerine karşılık olarak kaydedilen gerçek kan basıncı değerlerini kullanacak şekilde konfigure edilmiştir.The invention relates to a non-invasive blood pressure estimation system from ECG signals using machine learning (10). As an innovation, it includes an input unit (14) that receives instantaneous ECG signals from the user and a memory unit (11) that stores a machine learning model that calculates the estimated blood pressure value from the received ECG signal, and a processor unit (12) that transfers the results to a user interface. The machine learning model is configured to extract features from the real ECG signals and instantaneous ECG signals entered into the memory unit (11) using complexity analysis, process the extracted features using a stacking regression model, and use the recorded real blood pressure values corresponding to the real ECG signals when calculating the estimated blood pressure value corresponding to the instantaneous ECG signal.

Description

TARIFNAME KAN BASINCI TAHMIN SISTEMI TEKNIK ALAN Bulus, makine ögrenmesi ile EKG sinyallerinden invaziv olmayan kan basinci tahmin sistemi ile ilgilidir. ÖNCEKI TEKNIK Yaklasik üç yetiskinden biri yüksek tansiyona (hipertansiyon) sahiptir ancak çogu hasta bu durumun farkinda degildir. Yüksek tansiyon genellikle hiçbir uyari isareti vermeyen ancak kalp krizi veya felç gibi yasami tehdit eden durumlara yol açabilen oldukça önemli bir hastaliktir. Yüksek tansiyon kadar tehlikeli kabul edilmese de düsük tansiyon (hipotansiyon) da bas dönmesi, göz kararmasi, halsizlik ve bayginlik gibi sorunlara yol açtigi için dikkat edilmesi gereken bir durumdur. Tansiyon hastaliklarinin erken teshisi ve teshis sonucu yasam seklinde yapilacak degisiklikler tansiyonun insan sagligina zarar vermesini engelleyebilmektedir. Önemli ve önlenebilir olan tansiyon hastaliklarinin erken teshisi için kisilerin sürekli izlenebilmesi ve tespit edilen tansiyon sonuçlarinin elektrokardiyogram (EKG) gibi diger biyosinyaller ile birlikte degerlendirilmesi tansiyona bagli zararlarin en aza indirgenmesi için faydali olmaktadir. Nabiz, kan basinci, kan oksijen doygunlugu gibi biyosinyaller hayati sinyallerdir ve sürekli takipleri yasami tehdit edebilecek birçok sorunun ortadan kaldirilmasini saglayabilmektedir. Önlenebilir ölümlerin büyük bir çogunlugu kalp krizleri ve felçlerle iliskilidir. Kan basinci (KB) artisi, dünya çapinda kardiyovasküler hastalik ve erken ölümün önde gelen nedenlerindendir. Bununla birlikte, kan basincini izlemek için mevcut yöntemlerin çogu özel ekipman gerektirir ve sürekli takibe uygun kolayliklar barindirmamaktadir. EP2992820A2 numarali yayinda bir kan basinci ölçüm metodu açiklanmaktadir. Bahsedilen metod temel olarak PPG (plurality of photoplethysmogram) kullanimina dayanmaktadir. Sonuç olarak, yukarida bahsedilen tüm sorunlar, ilgili teknik alanda bir yenilik yapmayi zorunlu hale getirmistir. BULUSUN KISA AÇIKLAMASI Mevcut bulus yukarida bahsedilen dezavantajlari ortadan kaldirmak ve ilgili teknik alana yeni avantajlar getirmek üzere, bir kan basinci tahmin sistemi ile ilgilidir. Bulusun amaci, makine ögrenmesi ile EKG sinyallerinden invaziv olmayan kan basinci tahmin sistemi ortaya koymaktir. Yukarida bahsedilen ve asagidaki detayli anlatimdan ortaya çikacak tüm amaçlari gerçeklestirmek üzere mevcut bulus, makine ögrenmesi ile EKG sinyallerinden invaziv olmayan kan basinci tahmin sistemi ile ilgilidir. Buna göre; kullanicidan anlik EKG sinyallerinin alinmasini saglayan bir giris birimini ve alinan EKG sinyalinden tahmini kan basinci degerini hesaplayan bir makine ögrenmesi modelini saklayan bir hafiza biriminden aldigi sonuçlari bir kullanici arayüzüne aktaran bir islemci birimini içermesi ve de bahsedilen makine ögrenmesi modelinin; hafiza birimine girilmis gerçek EKG sinyallerinden ve anlik EKG sinyallerinden karmasiklik analizi ile sinyalin özniteliklerini çikartacak, çikartilan öznitelikleri istifleme regresyon modeli ile isleyecek ve anlik EKG sinyaline karsilik gelen tahmini kan basinci degerini hesaplarken bahsedilen gerçek EKG sinyallerine karsilik olarak kaydedilen gerçek kan basinci degerlerini kullanacak sekilde konfigure edilmis olmasidir. Bulusun mümkün bir yapilanmasinda, makine ögrenmesi modelinin; istifleme regresyon modelinin ilk asamasinda K-En Yakin Komsu, Destek Vektör, Ridge, Karar Agaci Regresyonu, AdaBoost ve XGoost regresyonlarindan en az iki tanesini içerecek sekilde konfigure edilmesidir. Bulusun mümkün bir yapilanmasinda, makine ögrenmesi modelinin; istifleme regresyon modelinin ilk asamasinda elde edilen tahmin degerlerini girdi olarak alan nihai bir regresyon modeli içerecek sekilde konfigure edilmesidir. Bulusun mümkün bir yapilanmasinda, makine ögrenmesi modelinin; istifleme regresyon modelinin ilk asamasinda elde edilen tahmin degerlerini girdi olarak alan nihai bir regresyon modelinin lasso regresyon modeli olmasidir. Bulusun mümkün bir yapilanmasinda, makine ögrenmesi modelinin; karmasiklik analizi ile sinyal hareketliligi, sinyal karmasikligi, sinyal entropisi, otokorelasyon katsayisi ve fraktal boyut özniteliklerinden en az birini çikartacak sekilde konfigure edilmis olmasidir. SEKILLERIN KISA AÇIKLAMASI Sekil 1' de bulus konusu kan basinci tahmin sisteminin temsili bir sematik görünümü verilmistir. BULUSUN DETAYLI AÇIKLAMASI Bu detayli açiklamada bulus konusu kan basinci tahmin sistemi sadece konunun daha iyi anlasilmasina yönelik hiçbir sinirlayici etki olusturmayacak örneklerle açiklanmaktadir. Sekil 1'e atfen bulus konusu kan basinci tahmin sistemi temel olarak kullanicidan anlik EKG (elektrokardiyogram) sinyallerinin alinmasini saglayan bir giris birimine, çesitli verilerin ve bir makine ögrenmesi modelinin saklandigi bir hafiza birimine, verileri isleyen bir islemci birimine ve tahmini kan basinci degerinin aktarildigi bir kullanici arayüzüne sahiptir. Bahsedilen hafiza biriminin içerisinde önceden profesyonel tibbi cihazlar ile alinmis gerçek EKG sinyalleri ve bu verilere karsilik gelen gerçek kan basinci degerleri bulunmaktadir. Bahsedilen makine ögrenmesi modelinde hem gerçek EKG sinyallerinin hem de anlik EKG sinyallerinin karmasiklik analizi ile öznitelikleri çikartilmaktadir. Belirlenen öz nitelikler sinyalin düzenliligi, düzensizligi, kaotikligi gibi soyut özelliklerine odaklanarak kalp fonksiyonlari hakkinda bilgi edinmeyi amaçlar. Karmasiklik analizi ile sinyal hareketliligi, sinyal karmasikligi, sinyal entropisi, otokorelasyon katsayisi ve fraktal boyut öznitelikleri olusturulmaktadir. Belirlenen öznitelikler, istifleme regresyon modeline girdi olarak kullanilmaktadir. Istifleme regresyon modelinde öncelikle veriler farkli regresyon modellerinde islenerek ara sonuçlar alinmaktadir. Sonrasinda bahsedilen ara sonuçlar nihai bir regresyon modelinde daha islenerek nihai tahmin elde edilmektedir. K-En Yakin Komsu, Destek Vektör, Ridge, Karar Agaci Regresyonu, AdaBoost ve XGoost regresyonlarindan en az iki tanesi istifleme modelinin ilk asamasinda kullanilan tahminleme modelleridir. Ara sonuçlarin islendigi Regresyon modeli olarak Lasso Regresyon modeli kullanilmaktadir. Bulus konusu kan basinci tahmin sistemine ait makine ögrenme modelinde ana girdi olarak EKG sinyallerinin öz niteliklerinin kullanilmasi ve tahminlemenin istifleme regresyon modeli ile yapilmasi sayesinde önceki teknige kiyasla çok daha dogru tahminlemeler yapilmasi saglanmaktadir. Bulusun koruma kapsami ekte verilen istemlerde belirtilmis olup kesinlikle bu detayli anlatimda örnekleme amaciyla anlatilanlarla sinirli tutulamaz. Zira teknikte uzman bir kisinin, bulusun ana temasindan ayrilmadan yukarida anlatilanlar isiginda benzer yapilanmalar ortaya koyabilecegi açiktir. TR TR TR TR TR TR TR TR TR TR TR TRDESCRIPTION BLOOD PRESSURE PREDICTION SYSTEM TECHNICAL FIELD The invention relates to a non-invasive blood pressure prediction system from ECG signals using machine learning. PREVIOUS TECHNIQUE Approximately one in three adults has high blood pressure (hypertension), but most patients are unaware of it. High blood pressure is a very serious condition that often gives no warning signs but can lead to life-threatening conditions such as heart attack or stroke. Although not considered as dangerous as high blood pressure, low blood pressure (hypotension) is also a condition that should be watched out for because it can cause problems such as dizziness, blurred vision, weakness, and fainting. Early diagnosis of blood pressure diseases and lifestyle changes resulting from diagnosis can prevent blood pressure from harming human health. For the early diagnosis of important and preventable hypertension, continuous monitoring of individuals and evaluation of detected blood pressure results together with other biosignals such as electrocardiogram (ECG) is beneficial in minimizing the harm caused by hypertension. Biosignals such as pulse, blood pressure, and blood oxygen saturation are vital signals, and their continuous monitoring can eliminate many life-threatening problems. A large majority of preventable deaths are related to heart attacks and strokes. Increased blood pressure (BP) is a leading cause of cardiovascular disease and premature death worldwide. However, most current methods for monitoring blood pressure require specialized equipment and do not offer the convenience of continuous monitoring. A blood pressure measurement method is described in publication EP2992820A2. The method mentioned is fundamentally based on the use of PPG (plurality of photoplethysmogram). Consequently, all the problems mentioned above have made it necessary to make an innovation in the relevant technical field. BRIEF DESCRIPTION OF THE INVENTION The present invention relates to a blood pressure prediction system aimed at eliminating the disadvantages mentioned above and bringing new advantages to the relevant technical field. The aim of the invention is to create a non-invasive blood pressure prediction system from ECG signals using machine learning. To achieve all the objectives mentioned above and detailed below, the present invention relates to a non-invasive blood pressure prediction system from ECG signals using machine learning. Accordingly; The invention includes an input unit that allows the user to receive instantaneous ECG signals, a memory unit that stores a machine learning model that calculates the estimated blood pressure value from the received ECG signal, and a processor unit that transmits the results to a user interface. Furthermore, the machine learning model is configured to extract features from both real and instantaneous ECG signals through complexity analysis, process these features using a stacking regression model, and use the recorded real blood pressure values corresponding to the real ECG signals when calculating the estimated blood pressure value corresponding to the instantaneous ECG signal. In a possible configuration of the invention, the machine learning model would: The first stage of the stacking regression model involves configuring it to include at least two of the following regression types: K-Nearest Neighbor, Support Vector, Ridge, Decision Tree Regression, AdaBoost, and XGoost. A possible configuration of the invention would involve configuring the machine learning model to include a final regression model that takes the predicted values obtained in the first stage of the stacking regression model as input. Another possible configuration would involve the machine learning model being a lasso regression model that takes the predicted values obtained in the first stage of the stacking regression model as input. Finally, a possible configuration would involve configuring the machine learning model to extract at least one of the following features through complexity analysis: signal mobility, signal complexity, signal entropy, autocorrelation coefficient, and fractal dimension. BRIEF DESCRIPTION OF THE FIGURES Figure 1 provides a representative schematic view of the blood pressure prediction system. DETAILED DESCRIPTION OF THE INVENTION This detailed description explains the blood pressure prediction system using examples that do not limit understanding and are intended to facilitate a better understanding of the subject. Referring to Figure 1, the blood pressure prediction system essentially consists of an input unit that receives real-time ECG (electrocardiogram) signals from the user, a memory unit that stores various data and a machine learning model, a processing unit that processes the data, and a user interface to which the predicted blood pressure value is transmitted. The memory unit contains real ECG signals previously obtained with professional medical devices and the corresponding real blood pressure values. In the aforementioned machine learning model, features are extracted from both real and instantaneous ECG signals through complexity analysis. The identified features focus on abstract characteristics such as the regularity, irregularity, and chaoticness of the signal, aiming to gain information about heart function. Complexity analysis generates features such as signal mobility, signal complexity, signal entropy, autocorrelation coefficient, and fractal dimension. The identified features are used as input to the stacking regression model. In the stacking regression model, the data is first processed in different regression models to obtain intermediate results. Subsequently, these intermediate results are further processed in a final regression model to obtain the final prediction. At least two of the following regressions are used in the first stage of the stacking model: K-Nearest Neighbor, Support Vector, Ridge, Decision Tree Regression, AdaBoost, and XGoost. The Lasso Regression model is used as the regression model for processing the intermediate results. The machine learning model for the blood pressure prediction system, which is the subject of this invention, uses the features of ECG signals as the main input and performs predictions using a stacking regression model, resulting in much more accurate predictions compared to the previous technique. The scope of patent protection is specified in the attached claims and cannot be limited to the examples given in this detailed description. It is clear that a technically skilled person could develop similar structures based on the above description without deviating from the main theme of the invention.

Claims (1)

1.1.
TR2023/009498A 2023-08-08 2023-08-08 BLOOD PRESSURE ESTIMATION SYSTEM TR2023009498A2 (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
TR2023/009498A TR2023009498A2 (en) 2023-08-08 2023-08-08 BLOOD PRESSURE ESTIMATION SYSTEM
PCT/TR2024/050889 WO2025034186A1 (en) 2023-08-08 2024-07-29 Blood pressure prediction system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TR2023/009498A TR2023009498A2 (en) 2023-08-08 2023-08-08 BLOOD PRESSURE ESTIMATION SYSTEM

Publications (1)

Publication Number Publication Date
TR2023009498A2 true TR2023009498A2 (en) 2023-08-21

Family

ID=94535143

Family Applications (1)

Application Number Title Priority Date Filing Date
TR2023/009498A TR2023009498A2 (en) 2023-08-08 2023-08-08 BLOOD PRESSURE ESTIMATION SYSTEM

Country Status (2)

Country Link
TR (1) TR2023009498A2 (en)
WO (1) WO2025034186A1 (en)

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8825428B2 (en) * 2010-11-30 2014-09-02 Neilcor Puritan Bennett Ireland Methods and systems for recalibrating a blood pressure monitor with memory
JP6913928B2 (en) * 2017-02-15 2021-08-04 国立大学法人 東京大学 Blood pressure measuring device, blood pressure measuring method and blood pressure measuring program
JP7511405B2 (en) * 2020-07-15 2024-07-05 日本光電工業株式会社 Apparatus and method for estimating biological parameter value, monitoring apparatus, and computer program
US11832919B2 (en) * 2020-12-18 2023-12-05 Movano Inc. Method for generating training data for use in monitoring the blood pressure of a person that utilizes a pulse wave signal generated from radio frequency scanning

Also Published As

Publication number Publication date
WO2025034186A1 (en) 2025-02-13

Similar Documents

Publication Publication Date Title
EP2956906B1 (en) Analysing video images of a subject to identify spatial image areas which contain periodic intensity variations
Jain et al. Face video based touchless blood pressure and heart rate estimation
US10004410B2 (en) System and methods for measuring physiological parameters
US9443289B2 (en) Compensating for motion induced artifacts in a physiological signal extracted from multiple videos
US8768438B2 (en) Determining cardiac arrhythmia from a video of a subject being monitored for cardiac function
US9436984B2 (en) Compensating for motion induced artifacts in a physiological signal extracted from a single video
CN105338890B (en) The method and apparatus for determining life parameters
Qiao et al. Revise: Remote vital signs measurement using smartphone camera
JP7065845B2 (en) Devices, systems, methods, and computer programs for obtaining vital signs of subjects
US9521954B2 (en) Video acquisition system for monitoring a subject for a desired physiological function
CN108135487A (en) For obtaining the equipment, system and method for the vital sign information of object
DE102013208587A1 (en) PROCESSING A VIDEO FOR VASCULAR PATTERN RECOGNITION AND HEART FUNCTIONAL ANALYSIS
Chong et al. Motion and noise artifact-resilient atrial fibrillation detection using a smartphone
JP2015211829A (en) Determination of arterial pulse wave transit time from vpg and ecg/ekg signal
AU2022374130B2 (en) Cardiac signal based biometric identification
Zeng et al. Infrared video based non-invasive heart rate measurement
US10779771B2 (en) Signal processing method and apparatus
Hernandez-de la Cruz et al. Simultaneous estimation of instantaneous heart and respiratory rates using image photoplethysmography on a single smartphone
KR102787507B1 (en) Method for measuring non-contact heart rate and computing device for executing the method
TR2023009498A2 (en) BLOOD PRESSURE ESTIMATION SYSTEM
Zhuang et al. Remote blood pressure measurement via spatiotemporal mapping of a short-time facial video
Bassiouni et al. Combination of ECG and PPG signals for smart healthcare systems: Techniques, applications, and challenges
Islam et al. A non-invasive technique of early heart diseases prediction from photoplethysmography signal
Shukla et al. Iot based non-invasive vital signs monitoring in neonatal intensive care unit (nicu)
Chugh et al. Effect of Different Signal Processing Techniques on a Calibration Free Pulse Oximeter