Chart-aware scanning
Google ML Kit is specified to find document edges, auto-crop the page, and correct photos taken at an angle before analysis begins.
Newborn growth monitoring
Turn paper growth charts into digital health insights.
NeoScan is designed to use computer vision and WHO growth standards to transform handwritten Road-to-Health charts into digital records, clear growth trends, and review flags.
Academic prototype
JKUAT · JHUB AFRICA
Kenya, 2026
01 Paper
02 Data
03 Insight
Designed for Kenya’s MCH contextBuilt around the MOH Road-to-Health chart format.
WHO growth standardsDay-precise LMS reference calculations.
Offline-first captureQueued scans for intermittent connectivity.
Consent-controlled sharingParents choose when a clinic receives access.
The problem
Important growth records often live on paper alone.
01
Records can be lost or damaged
The paper Road-to-Health booklet is carried between home and clinic throughout a child’s first five years.
02
Handwritten plots are hard to reuse
A visual mark on a chart is useful in person, but difficult to search, share, or compare over time.
03
Specialists may not see the full history
Remote review is difficult when records are only available inside a physical booklet.
04
Connectivity cannot be assumed
Community and clinic workflows need to keep working when mobile data is slow or unavailable.
The NeoScan approach
Keep the trusted paper workflow. Add a digital layer.
NeoScan is designed to read the chart parents and clinics already use—then make its measurements easier to verify, understand, and share.
1
Capture
Photograph the paper chart with a supported phone.
2
Correct
Detect boundaries, flatten perspective, and improve contrast.
3
Extract
Identify chart axes, curves, handwriting noise, and plotted points.
4
Verify
Show extracted values for a person to review before saving.
5
Understand
Plot trends against the relevant WHO reference curves.
Core capabilities
One careful workflow, from camera to clinical review.
WHO growth analytics
Measurements are designed to be matched to sex-specific, day-precise WHO LMS tables and expressed as Z-scores and percentile bands.
Trend visualization
Interactive SVG charts place a child’s verified measurements alongside the 3rd, 15th, 50th, 85th, and 97th reference curves.
Review flags, not diagnoses
Defined rules can flag a major percentile-band change or an impossible growth rate. Every alert requires interpretation by a qualified healthcare professional.
Offline capture and sync
Scans and metadata are designed to wait in encrypted local storage, then synchronize in chronological order when a connection returns.
Consent and auditability
Clinic access is designed around explicit, revocable parental consent, role boundaries, secure reports, and immutable activity records.
Product experience
See NeoScan in action.
Move through the proposed scan workflow. Screens show an illustrative product concept—not live patient data or validated results.
Check before saving
6 weeks4.6 kgDetected
10 weeks5.3 kgDetected
14 weeks6.0 kgReview
6 months6.4 kgDetected
Measurement placed against the relevant reference
Reference band: between the 15th and 50th percentiles
The production system is specified to calculate an exact Z-score from age-in-days, sex, and the relevant WHO L, M, and S values.
Weight-for-age
Data visualization
A growth history you can explore.
Select a point to see its date, measurement, and illustrative percentile band. The values shown are demo data and do not represent a real child.
Weight-for-age · child record
WHO reference curves Child measurements
Growth trend needs review
A significant change across percentile bands has been detected in this illustrative example. This is an automated flag and should be reviewed by a qualified healthcare professional.
Designed around real users
One record. Two clear experiences.
For parents and guardians
Simple steps and clear language.
Photograph the chart, check the values, see growth over time, and choose when to share a record with a clinic.
- Easy child profiles
- Plain-language growth information
- Offline scan capture
- Revocable sharing consent
For healthcare providers
Structured records with traceability.
Review extracted measurements, historical curves, anomaly flags, shared patient records, referrals, and audit history within defined role permissions.
- Assigned-record access
- Manual verification trail
- Secure PDF reporting
- Referral and review workflows
Planned immunization module
A schedule that shows what comes next.
The SRS specifies a milestone-based Kenyan immunization schedule with clear states. This interface concept presents that requirement without replacing clinic guidance.
Scope note: The SRS also lists immunization tracking as out of scope in its project summary. NeoScan therefore presents this as a planned requirement that needs scope confirmation, not a completed feature.
Birth milestone03 JanCompleted
6-week milestone14 FebCompleted
10-week milestone13 MarDue
14-week milestone10 AprUpcoming
Deferred itemClinic reviewDeferred
Completed records are specified to include administration date, batch number, clinician identity, and facility, with controlled corrections through the audit trail.
Privacy and security by design
Sensitive records need visible safeguards.
The project design references the Kenya Data Protection Act 2019 and Kenya Health Act. It does not claim legal certification or completed compliance validation.
Control stays close to the user.
Parents are specified to approve clinic sharing explicitly and can revoke access to future data.
Read the privacy design →EncryptedAES-256 at rest is a stated design requirement.
Secure connectionTLS 1.3 is specified for data in transit.
Role-basedUsers see records allowed by their role and consent.
AuditableCreate, read, update, share, and override actions are logged.
Technology
Mobile capture. Server-side vision. Structured health data.
The specified architecture combines on-device document correction with a FastAPI computer-vision pipeline and PostgreSQL-backed WHO reference tables.
Explore the architecture- React Native + Expo
Cross-platform parent and clinician application.
- Google ML Kit + OpenCV
Document detection, correction, normalization, and pen separation.
- YOLOv11 + FastAPI
Server-side chart element detection and inference orchestration.
- WHO LMS + PostgreSQL
Day-precise reference lookup, patient records, and audit trails.
Research and validation
Targets must be tested, not assumed.
NeoScan’s SRS is a draft specification dated June 2026. Accuracy, extraction tolerance, availability, and clinical workflow targets are requirements for validation—not published outcomes.
Planned methodology
Held-out chart testing
The proposed dataset uses volunteer-filled, non-patient MOH chart prints, varied devices and lighting, controlled degradation, and an 80/10/10 train-validation-test split.
Formal project deliverables
- Digitization Accuracy Report
Measures model detection and coordinate extraction against known chart values.
- Clinical Validation Study
Defines how clinical relevance and workflow fit should be evaluated.
- User Manual for Parents
Explains the experience in simple, accessible language.
- API Integration Guide
Documents future clinic and system integration pathways.
Frequently asked questions
Clear answers, without clinical overclaiming.
No. It is designed to digitize growth records, compare verified measurements with WHO growth references, and flag patterns that may need review. A qualified healthcare professional must interpret health concerns.
No. NeoScan adds a digital record to the existing paper workflow. The original card remains an important clinical document.
Yes. The specified workflow shows extracted values before saving. A parent or provider can correct a misread value, and that override is recorded in the audit trail.
The app is designed to save scans and child information in encrypted local storage, then process the queue when a connection returns.
The SRS defines 98% classification accuracy as a target for a held-out test set. It does not provide completed study results, so NeoScan does not present that target as an achieved result.
Parents are specified to see their own children’s records. Providers can see records explicitly shared with their clinic, within role-based permissions. Parents can revoke access to future data.
Project enquiries
Help shape a careful path from prototype to evidence.
Connect about research collaboration, clinical workflow review, chart digitization, or technical partnership.
Project updates
Follow the research journey.
This demonstration newsletter form validates locally and is ready for a future backend connection.