Validating algorithms for continuous blood pressure measurement in patients with sleep apnoea

Blood pressure derivation from short term raw PPG signals measured by a wearable device.

Status

Recruiting

Phases

Unknown

Study type

Observational

Source

ISRCTN

Registry ID

ISRCTN17666515

Enrollment

100

Registered

2019-06-04

Start date

2019-04-15

Completion date

Unknown

Last updated

2019-07-01

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Sleep apnoea Nervous System Diseases Sleep apnoea

Interventions

The algorithms will be first developed using an online database, i.e. Medical Information Mart for Intensive Care II (MIMIC II), which contains BP measurements using intra-arterial blood pressure, ECG

Eligibility

Sex/Gender

Female

Inclusion criteria

Inclusion criteria: Referred to the sleep lab for diagnostic polysomnography because of suspicion of sleep apnoea.

Exclusion criteria

Exclusion criteria: 1. Younger than 18 years. 2. Body Mass Index > 40 kg/m2. 3. Diagnosed with atrial fibrillation. 4. Have a pacemaker. 5. Dark skin tone.

Design outcomes

Primary

Measure	Time frame
The accuracy of blood pressure values estimated from wearable PPG is compared against values obtained using a blood pressure cuff at baseline during wake hours. The error (mean and standard deviation) between the estimated systolic BP/diastolic BP and the cuff value will be calculated. A T-test will evaluate significant differences. Bland-Altman analysis will be used to check the interchangeability of the two methods.	—

Measure

Time frame

The accuracy of blood pressure values estimated from wearable PPG is compared against values obtained using a blood pressure cuff at baseline during wake hours. The error (mean and standard deviation) between the estimated systolic BP/diastolic BP and the cuff value will be calculated. A T-test will evaluate significant differences. Bland-Altman analysis will be used to check the interchangeability of the two methods.

—

Secondary

Measure	Time frame
Parameters derived from polysomnography and wearable data during waking hours at baseline will be evaluated to differentiate patients with severe sleep apnea (AHI >= 30) from patients with milder or no sleep apnea (AHI < 30). In order to determine the best parameters that can differentiate patients with different AHI, two groups of patients will be created using the dataset collected for the first goal. One group will contain only patients with AHI < 30 and the other group will contain matched subjects suffering from patients with AHI = 30. Patients will be matched one-by-one using age, gender and BMI. This information is contained in the ‘LUCS: Vragenlijst’ questionnaire. For each computed parameter, the differences between the groups will be tested using the Wilcoxon test. In addition, apart from looking at significant levels (a=0.05), different classifiers based on linear discriminant analysis, Support Vector Machines (SVM), and Least-Squares SVM will be implemented for the separation of apnea patients with high and low AHI. Finally, the accuracy of the separation will be also analyzed using the F1 score.	—

Measure

Time frame

Parameters derived from polysomnography and wearable data during waking hours at baseline will be evaluated to differentiate patients with severe sleep apnea (AHI >= 30) from patients with milder or no sleep apnea (AHI < 30). In order to determine the best parameters that can differentiate patients with different AHI, two groups of patients will be created using the dataset collected for the first goal. One group will contain only patients with AHI < 30 and the other group will contain matched subjects suffering from patients with AHI = 30. Patients will be matched one-by-one using age, gender and BMI. This information is contained in the ‘LUCS: Vragenlijst’ questionnaire. For each computed parameter, the differences between the groups will be tested using the Wilcoxon test. In addition, apart from looking at significant levels (a=0.05), different classifiers based on linear discriminant analysis, Support Vector Machines (SVM), and Least-Squares SVM will be implemented for the separation of apnea patients with high and low AHI. Finally, the accuracy of the separation will be also analyzed using the F1 score.

—

Countries

Belgium

Contacts

Public ContactDorien Huysmans

dorien.huysmans@kuleuven.be016 37 92 69

Outcome results

None listed

Source: ISRCTN (via WHO ICTRP) · Data processed: Feb 4, 2026