Smartphone-based Non-Invasive Classification of Anemia Severity using Deep Learning: A Cross-Sectional Study in Rural India
- 1Google Research
Abstract
Anemia affects over two billion people globally. Screening individuals for anemia in community settings currently involves blood sampling, either a capillary or venous blood sample for running a point of care test or a laboratory grade blood test for hemoglobin, respectively. The need for a blood sample, and laboratory equipment and accessories are barriers to widespread screening for anemia in lower resource settings. We present a non-invasive, artificial intelligence (AI)-driven method for measuring hemoglobin (Hb) and detecting anemia using smartphone-captured images. We collected images of conjunctivae, as well as demographic and clinical data from patients seeking care at SEWA Rural hospital in rural, indigenous population of Gujarat, India. Ground truth Hb values were obtained using venous blood on a Sysmex XN-550 autoanalyser. Age and gender specific WHO reference cutoffs were used for classification of anemia severity. The dataset (n=5,176) was partitioned (70:15:15 ratio) for training (n = 3,623), validation (n = 776) and testing (n = 777). The deep learning based AI model used age, gender, pregnancy status and images of conjunctivae and predicted Hb values which were thresholded for various anemia classification tasks. Overall Hb prediction mean absolute error was 1.12 g/dL.The sensitivity and specificity of the model for classification of any anemia were 70% (95%CI 66-74) and 70% (95%CI 64-76) respectively. Sensitivity (90%, 95% CI 82-94) and specificity (80%, 95% CI 77-83) were higher for detecting severe anemia versus mild, moderate, or no anemia. The accuracy of the model was sustained in a variety of lighting conditions, skin tone and smartphone types. Our findings demonstrate the viability of smartphone based detection of anemia, especially severe anemia, in lower-resource settings and motivate further improvements and validation for use in community settings. The dataset comprising conjunctiva, tongue, nail bed images and smartphone-captured photoplethysmogram (PPG) signal is also released for community research use.
Related
- Modeling Factors Associated with Smoking Dependence Severity Using Binary Classification-Based Machine Learning: Prediction with Hematological and Demographic Patient Data
- Diffusion-augmented contrastive learning for diabetic retinopathy severity classification
- Machine learning-based association analysis and cross-sectional classification of physical exercise behaviors and mental health among vocational medical students
- The relationship between smartphone addiction, Mediterranean diet adherence, body composition, and quality of life among university students: a cross-sectional study
- Machine Learning with Social Determinants of Health for Cross-Sectional Diabetes Classification: Development and Temporal Validation.
BibTeX
@misc{desai2026smartphonebased,
title = {{Smartphone-based Non-Invasive Classification of Anemia Severity using Deep Learning: A Cross-Sectional Study in Rural India}},
author = {S. Desai and S. Gadgil and V. A. Gottala and K. Jesalpura and D. Ramnath and E. Gruzewska and H. Purandare and Y. Liu and U. Telang and J. A. Taylor and R. Sawhney and R. Tiwari and D. R. Webster and S. Shetty and A. Hassidim and Y. Matias and M. Howell and S. Venkatraman and B. Daryani and C. Chen and S. Kakarmath and M. Daswani and M. Wilson and R. Pullakhandam and S. Vispute and M. Thompson},
howpublished = {medRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.22.26366247},
url = {https://www.medrxiv.org/content/10.64898/2026.10.22.26366247v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Smartphone-based Non-Invasive Classification of Anemia Severity using Deep Learning: A Cross-Sectional Study in Rural India
AU - S. Desai
AU - S. Gadgil
AU - V. A. Gottala
AU - K. Jesalpura
AU - D. Ramnath
AU - E. Gruzewska
AU - H. Purandare
AU - Y. Liu
AU - U. Telang
AU - J. A. Taylor
AU - R. Sawhney
AU - R. Tiwari
AU - D. R. Webster
AU - S. Shetty
AU - A. Hassidim
AU - Y. Matias
AU - M. Howell
AU - S. Venkatraman
AU - B. Daryani
AU - C. Chen
AU - S. Kakarmath
AU - M. Daswani
AU - M. Wilson
AU - R. Pullakhandam
AU - S. Vispute
AU - M. Thompson
PY - 2026
DA - 2026/10/22
PB - medRxiv
DO - 10.64898/2026.10.22.26366247
UR - https://www.medrxiv.org/content/10.64898/2026.10.22.26366247v1
AB - Anemia affects over two billion people globally. Screening individuals for anemia in community settings currently involves blood sampling, either a capillary or venous blood sample for running a point of care test or a laboratory grade blood test for hemoglobin, respectively. The need for a blood sample, and laboratory equipment and accessories are barriers to widespread screening for anemia in lower resource settings. We present a non-invasive, artificial intelligence (AI)-driven method for measuring hemoglobin (Hb) and detecting anemia using smartphone-captured images. We collected images of conjunctivae, as well as demographic and clinical data from patients seeking care at SEWA Rural hospital in rural, indigenous population of Gujarat, India. Ground truth Hb values were obtained using venous blood on a Sysmex XN-550 autoanalyser. Age and gender specific WHO reference cutoffs were used for classification of anemia severity. The dataset (n=5,176) was partitioned (70:15:15 ratio) for training (n = 3,623), validation (n = 776) and testing (n = 777). The deep learning based AI model used age, gender, pregnancy status and images of conjunctivae and predicted Hb values which were thresholded for various anemia classification tasks. Overall Hb prediction mean absolute error was 1.12 g/dL.The sensitivity and specificity of the model for classification of any anemia were 70% (95%CI 66-74) and 70% (95%CI 64-76) respectively. Sensitivity (90%, 95% CI 82-94) and specificity (80%, 95% CI 77-83) were higher for detecting severe anemia versus mild, moderate, or no anemia. The accuracy of the model was sustained in a variety of lighting conditions, skin tone and smartphone types. Our findings demonstrate the viability of smartphone based detection of anemia, especially severe anemia, in lower-resource settings and motivate further improvements and validation for use in community settings. The dataset comprising conjunctiva, tongue, nail bed images and smartphone-captured photoplethysmogram (PPG) signal is also released for community research use.
ER -