Fingernail Photo Anaemia Screening App for Community Health Workers
Build an offline phone app that guides a community health worker to take a fingernail photo with a colour reference card, estimates whether the person is likely to be anaemic and tracks referral for a blood test. Anaemia is common among women and children in South Asia and often goes unnoticed because a blood test means travelling to a lab. The project reproduces a published smartphone method, checks it against a haemoglobin meter with clinical partners and is clear that it screens, not diagnoses.
Advanced
Needs a team, real planning, and a supervisor who knows the area.2 semesters · 3 students
7 to ship
4 optional extrasSuggested stack
What you should ship
- A guided capture screen with a printed colour reference card in frame, checks for blur, glare and nail polish, and a retake prompt
- Colour correction using the reference card so different phones and lighting give comparable values
- A model that estimates haemoglobin or anaemia risk from nail bed colour, trained on paired data collected with clinical partners
- An offline record per visit with a referral recommendation and follow-up reminders
- A sync step that uploads anonymised records when there is signal
- An evaluation report with sensitivity and specificity against a haemoglobin meter reading
- Ethics approval documents and a consent form in Urdu and English
If you have time left
- Comparing two or more phone models for consistency
- Adding inner eyelid photos as a second signal, with the same quality checks
- A supervisor dashboard of referrals and completed blood tests
- Running the full pipeline without any cloud step
The problem: anaemia that nobody measures
Anaemia, too little haemoglobin in the blood, causes tiredness, weakness and complications in pregnancy, and slows children's development. It is widespread among women and young children in South Asia. Diagnosis needs a blood test. For many families in rural areas, that means a trip to a lab, a cost and a day's lost work, so it is skipped until symptoms are serious.
Community health workers, such as Pakistan's Lady Health Workers, visit homes regularly. If they had a quick, non-invasive screen on a phone, they could send the people most likely to be anaemic for a proper test.
What already exists and the gap
- Haemoglobin meters used at clinics give accurate readings from a drop of blood but need consumables and training.
- Research apps estimate haemoglobin from smartphone photos of fingernails or the inner eyelid, with published studies in the United States and India.
- Health worker apps record visits and referrals but do not screen.
The gap is a screening app built for community health workers' real conditions: cheap phones, outdoor light, no signal, Urdu instructions, and a referral loop, validated honestly against blood tests in a local population.
Inspiration
"Smartphone app for non-invasive detection of anemia using only patient-sourced photos" by Robert Mannino and colleagues, published in Nature Communications in December 2018, estimated haemoglobin from fingernail bed colour in smartphone images and compared it with complete blood count results. A later study, "Smartphone-based point-of-care anemia screening in rural Bihar in India", published in Communications Medicine in 2023, tested the idea in a South Asian rural setting. Your project builds on both and focuses on capture quality and the referral loop.
What you build
- Community health workers take a guided photo, see a result of "refer for blood test" or "no referral needed", and record the visit offline.
- Supervisors see referrals and whether blood tests happened.
- Clinical partners get a validation dataset and a report they can review.
The app works fully offline, in Urdu and English, with large buttons and pictures showing how to hold the finger and the card. A result is never shown as a haemoglobin number to the person being screened, only as a referral decision, because a precise-looking number from a photo invites more trust than it deserves. Each referral creates a follow-up reminder for the next visit.
The technically interesting part
- Colour is fragile. Phone cameras process colour differently and lighting varies wildly. A printed reference card in every photo, and colour correction against it, is essential. Test how much it reduces differences between phones.
- Capture quality. Blur, glare, dirt, henna and nail polish ruin the signal. Detect them on the phone and ask for a retake rather than guess.
- Small data. You will have paired photo and haemoglobin data for perhaps a few hundred people. Keep the model simple, use cross-validation and report uncertainty.
- Screening thresholds. For screening, missing an anaemic person is worse than an extra referral. Choose thresholds with the clinical partner and show the trade-off.
How to evaluate it
- Sensitivity and specificity for anaemia against haemoglobin meter readings, with confidence intervals.
- Agreement between estimated and measured haemoglobin, shown as a Bland-Altman plot.
- Capture success: share of photos accepted first time by health workers.
- Usability: a short trial with health workers, recording time per screen and problems.
Suggested plan
- Weeks 1 to 5: read both studies, find a clinical partner, prepare the ethics application and consent forms.
- Weeks 6 to 10: capture screen, quality checks and colour correction, tested on team members.
- Weeks 11 to 14: data collection with the partner once approval is granted.
- Second semester, weeks 1 to 6: model, thresholds and on-phone inference.
- Weeks 7 to 10: referral records, sync and a usability trial.
- Weeks 11 to 14: evaluation, report and demo.
Path to a product
Public health programmes, NGOs and clinic chains need cheap screening tools that fit into home visits. A validated app could be offered to programmes that already equip health workers with phones. Even if accuracy is modest, a well-run validation study is valuable evidence and strong preparation for health technology roles.
Scope warning
This is a screening aid, never a diagnosis; the app must always say so and recommend a blood test. Do not collect any data from people until ethics approval is granted and consent is signed. Work only under the supervision of qualified clinicians. If ethics approval cannot be obtained in time, build and test the capture and colour pipeline and use published data only, and say so in the report.
Learn what you need on Sefism
- Mobile development roadmap: camera, offline storage and sync.
- Machine learning roadmap: small-data modelling and honest evaluation.
- Android Developers: Develop for Android: CameraX and on-device inference.
- OpenCV-Python Tutorials: colour correction and quality checks while prototyping.
- An Introduction to Statistical Learning: regression, cross-validation and uncertainty.
- Flutter Documentation: if you build the app in Flutter.
Get help with this idea
The risk is running out of time waiting for ethics approval or clinical data. Tauseef can help you plan the partnership early, structure the work so the app progresses meanwhile and frame the results responsibly. If you are new here, create a free account first.
Ideas and guidance, not finished projects
These are project ideas and scoping guidance, published free for students to use as a starting point. I do not build, write, or sell final-year projects, and I do not complete coursework for anyone. Take an idea, make it yours, and build it.
Not sure this one fits you?
I help students pick an idea that matches their skills and their deadline, then scope it down until it is finishable. Guidance only, never done for you.
Stuck on something specific?
Writing only gets you so far. If you want an answer to your situation rather than the general case, book a session and we will work through it together. Sessions are free for approved Sefism members, and a few slots open each week.
Follow along
New writing, resources and project ideas land here first.