# Feasibility of a Noncontact Photoplethysmography–Based Mobile App for Noninvasive Hemoglobin Monitoring: Exploratory Observational Study

## TL;DR

A mobile app using a smartphone camera to estimate hemoglobin levels was tested and showed moderate accuracy compared to lab results.

## Contribution

This study explores a novel noncontact, smartphone-based method for noninvasive hemoglobin monitoring.

## Key findings

- The app achieved a mean absolute error of 1.46 g/dL and a mean absolute percentage error of 11.26%.
- The Bland-Altman plot showed a mean difference of −0.70 g/dL between app and lab measurements.
- The app demonstrated an overall accuracy of 75% and an area under the curve of 0.701.

## Abstract

Anemia is a widespread global health issue. Hemoglobin (Hb) concentration measurement remains the most common method for anemia screening and diagnosis. In recent years, there has been growing interest in the development of noninvasive point-of-care technologies that eliminate the need for blood sampling.

This pilot study explores the feasibility of using a noncontact photoplethysmography–based mobile app for Hb monitoring.

Adult volunteers aged 18 years and older, of both sexes, were consecutively recruited. Participants were seated and allowed a 2-minute rest before measurements. During testing, they faced a smartphone running comestai.app, which used the front-facing camera to capture facial videos. Simultaneous readings were collected for Hb over approximately 90 seconds using the app. Ambient lighting was standardized for all remote photoplethysmography recordings. No medical decisions were made based on the app-generated data. A complete blood count, including Hb levels, was used as a reference for comparison with the data collected using comestai.app.

A total of 555 (female: n=313, 56.4%; male: n=242, 43.6%) individuals participated in the study. The app achieved a mean absolute error of 1.46, a mean absolute percentage error of 11.26, a mean error of −0.67, and a root mean square error of 1.88. The Bland-Altman plot evaluated the agreement between the app-based and laboratory-based Hb measurements, with the mean difference between the 2 methods being −0.70 g/dL. The method demonstrated an overall accuracy of 75%. The area under the curve was 0.701 (95% CI 0.655-0.745).

Comestai.app offers an innovative approach to wellness monitoring by providing noninvasive Hb estimation using the smartphone’s front-facing camera. Continued development, including algorithmic refinement and larger-scale validation in diverse populations, will be key to enhancing accuracy and broadening its utility. By leveraging the ubiquity of smartphones, comestai.app contributes meaningfully to the democratization of health monitoring and the promotion of proactive self-care.