Development and Multicenter Validation of an AI-Based Remote Photoplethysmography (rPPG) Facial Scan for Multimodal Health Assessment

Official title

Development and Multicenter Validation of an AI-Based Remote Photoplethysmography (rPPG) Facial Scan for Multimodal Health Assessment: Agreement With Clinical, Laboratory, and Psychological Parameters in an Urban Population

Overview

The goal of this observational study is to learn if a non-contact facial scan using artificial intelligence (AI) can be used to check health status in adults living in urban areas such as Jakarta. The facial scan uses a method called remote photoplethysmography (rPPG), which measures small changes in blood flow from the face using a camera. The main questions this study aims to answer are: 1. How close are the results from the facial scan to standard medical measurements, such as heart rate, breathing rate, blood pressure, and oxygen levels? 2. Can the facial scan estimate other health indicators, such as blood sugar, lipid profile, HbA1c, and hemoglobin levels? 3. Is there a relationship between the facial scan results and mental health, such as stress, anxiety, and depression? Participants will take part in several simple and mostly non-invasive procedures: 1. Answer questionnaires about their mental health and daily habits 2. Have basic health checks, such as blood pressure, heart rate, and body measurements 3. Provide a blood sample for laboratory testing 4. Complete a facial scan using a camera for about 1 to 3 minutes Researchers will compare the results from the facial scan with standard clinical and laboratory tests to see how well the technology works. This study may help develop a simple and accessible screening tool that can be used for early detection of health risks. It may also support the use of digital health and telemedicine in community and clinical settings.

Detailed description

Remote photoplethysmography (rPPG) is an emerging non-contact optical technology that enables extraction of physiological signals from facial video using standard cameras. This approach has gained increasing attention in telemedicine due to its scalability, cost-effectiveness, and ability to perform remote health screening. Recent advancements in artificial intelligence (AI) have further expanded the potential of rPPG beyond basic vital sign monitoring to include estimation of cardiometabolic biomarkers and health risk indices. However, comprehensive validation of rPPG-based systems against standardized clinical measurements, laboratory biomarkers, and psychological parameters remains limited, particularly in low- and middle-income settings such as Indonesia. Given the high burden of cardiometabolic diseases in urban populations like Jakarta, evaluating the accuracy and feasibility of AI-based facial scanning technologies is essential to support early detection and digital health integration.

Specific Objectives

1. To assess the agreement between rPPG derived vital signs (heart rate, respiratory rate, blood pressure, SpO₂) and corresponding measurements obtained from standardized physical examination by trained personnel and validated medical devices 2. To determine the degree of concordance between rPPG based estimates and laboratory values of hemoglobin, blood glucose, HbA1c, LDL, HDL, triglycerides, and total cholesterol. 3. To analyze the association between rPPG derived physiological parameters and levels of depression, anxiety, and stress as measured by the DASS 21 questionnaire. 4. To calculate mean arterial pressure (MAP), ASCVD risk scores, and heart age from rPPG outputs and to compare these indices with those derived from standard clinical and laboratory data. 5. To develop and preliminarily evaluate exploratory algorithms using rPPG video data to estimate kidney function, liver function, muscle mass, visceral fat, body weight, body height, body mass index, and subcutaneous fat as potential screening parameters.

Methods This study will employ a multicenter observational design conducted across selected subdistricts in Jakarta and expanded to the Jabodetabek region. Adult participants will undergo comprehensive assessment including psychological questionnaires (DASS, PHQ, GAD), anthropometric measurements, body composition analysis, spirometry, muscle strength testing, and venous blood sampling. Blood samples will be analyzed using POCT (≤30 minutes) and ISO-standardized clinical laboratory methods. In parallel, participants will undergo a non-contact facial scan, generating rPPG-based outputs including vital signs, hemodynamic indices, and AI-estimated biomarkers. Statistical analysis will include Bland-Altman agreement analysis, Cohen's kappa for categorical variables, correlation analysis, and machine learning performance metrics (MAE, MSE, RMSE, R²).

Expected Results It is expected that rPPG-based measurements will demonstrate good agreement with standard clinical measurements for core vital signs (heart rate, respiratory rate, SpO₂), with moderate agreement for blood pressure and selected biomarkers. AI-based models are anticipated to show acceptable predictive performance for certain metabolic parameters and exploratory variables, supporting the feasibility of rPPG as a screening tool. The study is also expected to identify key confounding factors, such as skin tone and demographic variability, influencing signal accuracy.

Primary outcome measures

• Agreement of rPPG-Derived Vital Signs With Standardized Clinical Measurements [Time frame: At a single study visit during baseline assessment (cross-sectional measurement)]
• Concordance Between rPPG-Derived Biomarker Estimates and Standard Laboratory Measurements [Time frame: At a single study visit during baseline assessment (cross-sectional measurement)]
• Association Between rPPG-Derived Physiological Parameters and Psychological Status [Time frame: At a single study visit during baseline assessment (cross-sectional measurement)]
• Agreement of rPPG-Derived Cardiovascular Risk Indices With Standard Clinical Calculations [Time frame: At a single study visit during baseline assessment (cross-sectional measurement)]

Secondary outcome measures (1)

• Predictive Performance of rPPG-Based Models for Estimation of Organ Function and Body Composition [Time frame: At a single study visit during baseline assessment (cross-sectional measurement)]

Eligibility criteria

Inclusion criteria

• Adults aged ≥18 years.
• Able and willing to provide written informed consent.
• Able to comply with study procedures, including face scan, physical examination, blood sampling, and questionnaire completion.
• Clinically stable at the time of assessment.

Exclusion criteria

• Facial conditions affecting the region of interest (ROI), such as injury, deformity, or impaired circulation, that may interfere with rPPG signal acquisition.
• Presence of facial tattoos or coverings that obstruct optical signal detection.
• Inability to remain still or comply with measurement procedures during data acquisition.
• Severe medical conditions that preclude safe participation, as judged by the investigator.
• Incomplete data or withdrawal of consent during the study.

\-

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: Yes

Study design

Observational model

Ecologic or community

Study locations

Center list to be confirmed — check the primary protocol.

Publications

• Tan SYL, Chai JX, Choi M, Javaid U, Tan BPY, Chow BSY, Abdullah HR. Remote Photoplethysmography Technology for Blood Pressure and Hemoglobin Level Assessment in the Preoperative Assessment Setting: Algorithm Development Study. JMIR Form Res. 2025 Jun 6;9:e60455. doi: 10.2196/60455. PMID 40479628
• Ahmad Hatib NA, Lee JH, Chong SL, Sng QW, Tan VSR, Ong GY, Lim AM, Quek BH, How MS, Chan JMF, Saffari SE, Ng KC. A two-phased study on the use of remote photoplethysmography (rPPG) in paediatric care. Ann Transl Med. 2024 Jun 10;12(3):46. doi: 10.21037/atm-23-1896. Epub 2024 May 27. PMID 38911566
• Allado E, Poussel M, Renno J, Moussu A, Hily O, Temperelli M, Albuisson E, Chenuel B. Remote Photoplethysmography Is an Accurate Method to Remotely Measure Respiratory Rate: A Hospital-Based Trial. J Clin Med. 2022 Jun 24;11(13):3647. doi: 10.3390/jcm11133647. PMID 35806932
• Padaki AS, Zarzour AL, Keene KR, Canepa CA, Levin DR, Antonsen EL. Clinical validation of non-contact vital signs in an emergency department setting. Front Med Technol. 2026 Jan 20;7:1728913. doi: 10.3389/fmedt.2025.1728913. eCollection 2025. PMID 41640807
• Brown A, Tulkens J, Mattelin M, Sanglet T, Dhuyvetters B. Remote photoplethysmography for health assessment: a review informed by IntelliProve technology. Front Digit Health. 2026 Jan 5;7:1667423. doi: 10.3389/fdgth.2025.1667423. eCollection 2025. PMID 41561164
• Heiden E, Jones T, Brogaard Maczka A, Kapoor M, Chauhan M, Wiffen L, Barham H, Holland J, Saxena M, Wegerif S, Brown T, Lomax M, Massey H, Rostami S, Pearce L, Chauhan A. Measurement of Vital Signs Using Lifelight Remote Photoplethysmography: Results of the VISION-D and VISION-V Observational Studies. JMIR Form Res. 2022 Nov 14;6(11):e36340. doi: 10.2196/36340. PMID 36374541
• Debnath U, Kim S. A comprehensive review of heart rate measurement using remote photoplethysmography and deep learning. Biomed Eng Online. 2025 Jun 20;24(1):73. doi: 10.1186/s12938-025-01405-5. PMID 40542336
• Wiffen L, Brown T, Brogaard Maczka A, Kapoor M, Pearce L, Chauhan M, Chauhan AJ, Saxena M; Lifelight Trials Group. Measurement of Vital Signs by Lifelight Software in Comparison to Standard of Care Multisite Development (VISION-MD): Protocol for an Observational Study. JMIR Res Protoc. 2023 Jan 11;12:e41533. doi: 10.2196/41533. PMID 36630158

Identifiers

NCT: NCT07491978 · 20260319

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗