Non-invasive estimate of blood glucose and blood pressure from a photoplethysmograph by means of machine learning techniques
Open accessAbstract
Objective This work presents a system for a simultaneous non-invasive estimate of the blood glucose level (BGL) and the systolic (SBP) and diastolic (DBP) blood pressure, using a photoplethysmograph (PPG) and machine learning techniques. The method is independent of the person whose values are being measured and does not need calibration over time or subjects.Methodology The architecture of the system consists of a photoplethysmograph sensor, an activity detection module, a signal processing module that extracts features from the PPG waveform, and a machine learning algorithm that estimates the SBP, DBP and BGL values. The idea that underlies the system is that there is functional relationship between the shape of the PPG waveform and the blood pressure and glucose levels.Results As described in this paper we tested this method on 410 individuals without performing any personalized calibration. The results were computed after cross validation. The machine learning techniques tested were: ridge linear regression, a multilayer perceptron neural network, support vector machines and random forests. The best results were obtained with the random forest technique. In the case of blood pressure, the resulting coefficients of determination for reference vs. prediction were R(SBP)(2)=0.91, R(DBP)(2)=0.89, and R(BGL)(2)=0.90. For the glucose estimation, distribution of the points on a Clarke error grid placed 87.7% of points in zone A, 10.3% in zone B, and 1.9% in zone D. Blood pressure values complied with the grade B protocol of the British Hypertension society.Conclusion An effective system for estimate of blood glucose and blood pressure from a photoplethysmograph is presented. The main advantage of the system is that for clinical use it complies with the grade B protocol of the British Hypertension society for the blood pressure and only in 1.9% of the cases did not detect hypoglycemia or hyperglycemia.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Prior works developed approaches from at most 100 individuals and in most cases recorded less than a day of data.
“These works developed approaches from at most 100 individuals and in most cases, recorded less than a day of data.”
From Sleep and temperature data from wearable devices support noninvasive detection of diabetes mellitus in a large-scale, retrospective analysis (2026) · cites this paper as [19]
Blood glucose level has been estimated from PPG pulse wave shape.
“Blood glucose level, widely used for diabetes self-management, has been estimated from PPG pulse wave shape”
From Wearable Photoplethysmography for Cardiovascular Monitoring (2022) · cites this paper as [152]
In a population of 134 healthy, Caucasian subjects (median age 43, total range: 13-72), there was a statistically significant decrease in PTT at the toes (r2 = 0.48), fingers (r2 = 0.26), and ears (r2 = 0.15).
“By analyzing the change in PTT as a function of age within a population of 134 healthy, Caucasian subjects (median age 43, total range: 13–72), it was found that there is a statistically significant decrease at the toes (r2 = 0.48), fingers (r2 = 0.26), and ears (r2 = 0.15) [91,92].”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
As age increases, PTT decreases.
“PTT is inversely related to PWV—the same authors note in a different work that as age increases, PTT decreases.”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
Artificial intelligence and machine learning have been used to determine blood pressure from PPG and mitigate the effects age have on the estimate.
“Artificial intelligence and machine learning have been used to determine blood pressure from PPG and mitigate the effects age have on the estimate.”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
Monte-Moreno's work correlated with both systolic and diastolic blood pressure with an r2 of approximately 0.90.
“While this work utilized a population of individuals with good cardiovascular health, it was found to correlate with both systolic and diastolic blood pressure, with an r2 of approximately 0.90 [92].”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
An algorithm that incorporates age into systolic and diastolic blood pressure determination was created.
“An algorithm that incorporates age into systolic and diastolic blood pressure determination was created.”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
Monte-Moreno utilized machine learning to determine blood pressure from a PPG obtained from subjects ranging from 9-80 years old.
“In 2010, Monte-Moreno utilized machine learning to determine blood pressure from a PPG obtained from subjects ranging from 9–80 years old.”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
PTT decreases as age increases.
“| PTT | Decrease | Allen et al., 2002 Monte-Moreno, 2011 | [91,92] |”
From Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring (2021) · cites this paper as [92]
The citation lists haven’t been fetched yet
References 51 · Cited by 311
- Authors
- Monte-Moreno, E.
- Affiliations
- Universitat Politècnica de Catalunya (ES)
- Corresponding author
- Monte-Moreno, Enric
- Venue
- Artificial Intelligence in Medicine
- Publisher
- Elsevier BV
- Published
- 2011-10
- Volume
- 53
- Issue
- 2
- Pages
- 127-138
- ISSN
- 0933-3657 (ISSN-L)
- OpenAlex
- W1985034572
- Cited by
- 311
- References
- 51
- Citations by year
- 2026: 24 · 2025: 32 · 2024: 36 · 2023: 26 · 2022: 35 · 2021: 32 · 2020: 31 · 2019: 26 · 2018: 22 · 2017: 13 · 2016: 16 · 2015: 10 · 2014: 4 · 2013: 3 · 2012: 1
- FWCI
- 1.51
- Type
- journal article
- Language
- en
- Open access
- green
- hdl.handle.net
- Licence
- Locations
- Artificial Intelligence in Medicine (publishedVersion)
- PubMed (publishedVersion)
- RECERCAT (Consorci de Serveis Universitaris de Catalunya) (acceptedVersion, other-oa)
- UPCommons institutional repository (Universitat Politècnica de Catalunya) (submittedVersion, public-domain)
- Topics
- Physical Sciences › Engineering › Biomedical Engineering › Non-Invasive Vital Sign Monitoring
- Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Cardiovascular Health and Disease Prevention
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- blood pressure estimation, photoplethysmography, random forest, activity detection, Clarke error grid
- MeSH terms
- Algorithms, Artificial Intelligence, Blood Glucose, Blood Pressure, Blood Pressure Determination, Humans, Hypertension, Linear Models, Photoplethysmography
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
