Skip to contentMasud Rabbani
CV
Menu

Research focus

Everyday devices as health sensors.

Explore Masud Rabbani’s research on smartphone-based, non-invasive physiological sensing for blood cells, glucose, oxygen saturation, blood pressure, and multi-parameter mHealth.

13 related publications · 1 project

Research overview

Non-Invasive Health Sensing with Smartphones and AI

My non-invasive health-sensing research investigates whether smartphones and other accessible devices can estimate physiological measures that usually require specialized equipment, contact sensors, or laboratory procedures. The portfolio includes studies of white blood cells, glucose and HbA1c, oxygen saturation, blood pressure, hemoglobin, and multi-parameter monitoring. These projects combine optical sensing, camera video, photoplethysmography, signal-quality controls, and machine learning within mobile-health workflows.

The motivation is access. A measurement that can be collected with a familiar device may support more frequent monitoring, faster screening research, and care outside a centralized laboratory. But accessibility is only useful when measurement quality is visible and uncertainty is handled honestly. Smartphone cameras, illumination, skin-device contact, motion, ambient light, hardware variation, and population differences can all alter the observed signal. For that reason, the research program treats acquisition design and validation as core computational problems—not as details to address after a model is built.

Published work in this area includes a smartphone-based white-blood-cell counter using blue light and a static magnetic field, camera-based remote photoplethysmography for contactless oxygen-saturation monitoring, facial-video analysis for blood-pressure estimation, and quantum-machine-learning experiments for glucose and HbA1c estimation. A broader non-invasive mHealth platform and the dDream program connect individual measurements into a scalable multi-parameter architecture. Together, these studies explore what is technically possible while identifying the evidence still needed before a prototype could influence clinical decisions.

My contribution

How I contributed to this research program.

My documented contributions across these coauthored smartphone-sensing studies include conceptualization, validation, formal analysis, visualization, and scientific writing. The PulseSight author statement credits me with conceptualization, validation, visualization, and original-draft writing; the glucose and HbA1c paper credits me with formal analysis and review/editing; UbiHeart credits me with validation and review/editing; and UbiWhite credits me with reviewing and proofreading the article.

Together, these projects investigate how camera video and photoplethysmography can support research on white-blood-cell counts, glucose and HbA1c, oxygen saturation, and blood pressure. I present their results as early research evidence: their limited cohorts and acquisition settings do not establish that a smartphone can replace laboratory or clinically validated measurements.

Key findings

Evidence from the published studies.

Each result links to its complete publication record, abstract, authors, affiliations, and publisher source.

White-blood-cell estimation

UbiWhite evaluated 20 adults against CBC reference measurements and reported RMSE values of 1.04 and 1.23 for its two algorithms.

Read the supporting paper

Glucose and HbA1c study

The study used ten-second fingertip videos from 136 participants and extracted 45 PPG features. Within its dataset, the quantum SVM reported 89.30% accuracy for glucose and 96.30% for HbA1c, with a 92.50% average cross-validation score for both.

Read the supporting paper

Contactless oxygen saturation

PulseSight evaluated 40 facial-video sets from 10 participants under different illumination conditions and reported RMSE of 1.45 ± 0.1, MAE of 0.92 ± 0.01, and a System Usability Scale score of 80.5.

Read the supporting paper

Questions

The problems guiding this research.

Signal quality before prediction

Can a mobile system recognize poor illumination, motion, weak optical coupling, and other acquisition failures before producing an estimate that appears more certain than the data allow?

Generalization across people and phones

How do models behave across skin characteristics, health conditions, camera sensors, operating systems, and collection environments that were not represented in the original study?

Useful multi-parameter monitoring

Which measurements can share a practical acquisition workflow, and how should a mobile platform communicate trends, confidence, and the need for confirmatory testing?

Research method

From a measurable signal to a defensible result.

  1. Design the acquisition

    Select camera, light, contact, facial-video, or PPG configurations around the physiological mechanism being studied and document the conditions that affect signal quality.

  2. Control the signal

    Detect motion and lighting artifacts, normalize device-dependent measurements, and extract physiological representations before asking a model to predict an outcome.

  3. Model and benchmark

    Compare classical, deep-learning, and quantum-machine-learning approaches against appropriate ground truth rather than reporting one favorable model in isolation.

  4. Test translation limits

    Evaluate repeatability, subgroup performance, device variation, usability, and calibration needs so the boundary between research prototype and clinical tool is explicit.

Related systems

Projects carrying the research into practice.

All projects

EyeGaze

An mHealth Approach to the Clinical Assessment of Gaze Stability and Telerehabilitation of Vestibular Dysfunction Funded by the National Institute on Deafness and Other Communication Disorders (Grant Number: 1R41DC022209-01)

Selected scholarship

Read the evidence behind this research focus.

All publications

Journal article · 2026

Toward a Noninvasive mHealth Platform

Nafi Us Sabbir Sabith, Sayed Mashroor Mamun, Masud Rabbani, Sheikh Iqbal Ahamed. “Toward a Noninvasive mHealth Platform”. Computer, 59(1), 95–107. (2026). https://doi.org/10.1109/MC.2025.3600446.

Read

Journal article · 2025

PulseSight: A novel method for contactless oxygen saturation (SpO2) monitoring using smartphone cameras, remote photoplethysmography and machine learning

Kazi Zawad Arefin, Kazi Shafiul Alam, Sayed Mashroor Mamun, Nafi Us Sabbir Sabith, Masud Rabbani, Parama Sridevi, Sheikh Iqbal Ahamed. “PulseSight: A novel method for contactless oxygen saturation (SpO2) monitoring using smartphone cameras, remote photoplethysmography and machine learning”. Smart Health, 36, 100542. (2025). https://doi.org/10.1016/j.smhl.2025.100542.

Read

Conference paper · 2026

dDream: A Smartphone-Based Comprehensive and Scalable Multi-Parameter Physiological Monitoring Platform

Sayed Mashroor Mamun, Kazi Shafiul Alam, Nafi Us Sabbir Sabith, Kazi Zawad Arefin, Masud Rabbani, Sheikh Iqbal Ahamed. “dDream: A Smartphone-Based Comprehensive and Scalable Multi-Parameter Physiological Monitoring Platform”. 2026 IEEE 50th Annual Computers, Software, and Applications Conference (COMPSAC), 3058–3067. (2026). https://doi.org/10.1109/COMPSAC69091.2026.00457.

Read

Conference paper · 2025

A Survey on Non-Invasive Computing: Neurological-Hematological Framework for Early Infection and Stroke Detection with Future Directions

Masud Rabbani, Nafi Us Sabbir Sabith, Sheikh Iqbal Ahamed. “A Survey on Non-Invasive Computing: Neurological-Hematological Framework for Early Infection and Stroke Detection with Future Directions”. 2025 IEEE 49th Annual Computers, Software, and Applications Conference (COMPSAC), 824–833. (2025). https://doi.org/10.1109/COMPSAC65507.2025.00110.

Read

Conference paper · 2026

FedHemo: A Federated CTAB-GAN+ Framework for Privacy-Preserving Non-Invasive Hemoglobin Prediction using Smartphone-based PPG Signals

Nafi Us Sabbir Sabith, Nathaniel Parise, Sachin Shrestha, Tasnia Zaman, Iysa Iqbal, Anushka Kolli, Masud Rabbani, Sheikh Iqbal Ahamed. “FedHemo: A Federated CTAB-GAN+ Framework for Privacy-Preserving Non-Invasive Hemoglobin Prediction using Smartphone-based PPG Signals”. 2026 IEEE 14th International Conference on Healthcare Informatics (ICHI), 1539–1547. (2026). https://doi.org/10.1109/ICHI69079.2026.00248.

Read

Conference paper · 2022

Towards a Survey on Universal Human Vital Signs with prototype for Detection and Record Electronically Acceptable Medical-data (dDream)

Masud Rabbani, Kazi Shafiul Alam, Lin He, Shiyu Tian, Mohammad Syam, Iysa Iqbal, Anushka Kolli, Hansika Kolli, Syeda Shefa, Bipasha Sobhani, Paramita Basak Upama, Sheikh Iqbal Ahamed. “Towards a Survey on Universal Human Vital Signs with prototype for Detection and Record Electronically Acceptable Medical-data (dDream)”. 2022 IEEE 46th Annual Computers, Software, and Applications Conference (COMPSAC), 502–511. (2022). https://doi.org/10.1109/COMPSAC54236.2022.00094.

Read

Responsible translation

Promising research still needs careful validation.

The long-term goal is not to replace established clinical testing with an unverified phone estimate. It is to determine where accessible sensing can support research, screening, longitudinal context, or a prompt for confirmatory care. A useful system must make failed acquisition and uncertainty as visible as a successful result.

Moving from a promising study to trustworthy use requires prospective evaluation, diverse participants, comparisons against recognized reference measurements, external replication, privacy protection, and regulatory review where appropriate. These requirements guide how the work is described and which claims are intentionally avoided.

Continue exploring

Connected areas of the research program.

Research focusEEG & brain-computer interfacesResearch focusMobile-health AIResearch focusAutism care & mCAREResearch focusConversational AI & VoiSResearch focusQuantum ML for healthcareResearch focusBangla NLP & language technology