EyeModelHub
Home› Subspecialties› Oculoplastics› Model card
§Model Card

Smartphone Eyelid Measurement AI (MAIA)

Automated regression-based measurement of MRD1, MRD2, and levator function (LF) from smartphone photographs

Chen H.C. · 2021 · Taiwan · doi:10.2196/32444
Open DOI
Measurement/Regression CNN External photo Internally validated
▸ More attributes
Anatomy · Periorbital soft tissueConditions · BlepharoptosisSingle-centern = 411 patients · 822 eyes images
§01Overview

Brief overview

This smartphone-based AI system uses multiple convolutional neural network models within proprietary MAIA software to automatically predict eyelid measurements (MRD1, MRD2, and levator function) from standardized orbital photographs captured on an iPhone. Across 822 eyes from 411 adults at a single Taiwanese center, the MRD1 and MRD2 models achieved excellent agreement with clinician-derived gold-standard measurements (ICC 0.90 and 0.84; MAE 0.35 mm and 0.37 mm), while levator function prediction showed more modest performance (ICC 0.69; MAE 1.06 mm). The model was internally validated using held-out test data only, with no external validation, public code, pretrained weights, or clinical deployment reported

§02Performance
Headline performance

MRD1: MAE 0.35 mm; ICC 0.90.

Full metrics — as reported
MRD1: MAE 0.35 mm; ICC 0.90.
MRD2: MAE 0.37 mm; ICC 0.84.
LF: MAE 1.06 mm; ICC 0.69.
Processing time ~2 seconds per set.
Calibration · Uncertainty
Not reported.
Bias · Fairness
Not reported.
§03Data & Validation
Study designDiagnostic accuracy study
Center typeSingle-center
Patients411
Images / eyes822 eyes
Split90/10 split + internal 80/20 train-validation subdivision + single-center hold-out test set
Reporting frameworks None stated
Age range>18
Sex (% female)Not reported
Race / ethnicity reportedN
Subgroup performanceN
§04Deployment
Deployment maturity
  1. Concept
  2. Prototype
  3. Internal val.
  4. External val.
  5. Regulator
  6. Clinical
Regulatory status
None No FDA / Health Canada / CE clearance.
Artifacts & license
○Code repository none
○Pretrained weights none
○Live demo none
License Proprietary
§05Quality Review · APPRAISE-AI
Total score
40 /100 Moderate
ClinicalDataMethodRobust.Report.Repro.
Sub-domain scores
Clinical relevance 4/4 100%
Data quality 12/24 50%
Methodological conduct 4/20 20%
Robustness of results 5/20 25%
Reporting quality 12/12 100%
Reproducibility 3/20 15%
▸ Why this score?

The APPRAISE-AI tool grades AI/ML medical studies across 24 items in 6 subdomains, scored independently by two reviewers. The total reflects the methodological transparency of the publication — not the underlying clinical value of the model.

How to read this: high scores in clinical relevance and reporting with low scores in methodological conduct, robustness, or reproducibility is the dominant pattern in ophthalmic AI today.

Quality bands: Low <40 · Moderate 40–59 · High 60–79 · Very High 80+. Most published ophthalmic AI sits in the Low–Moderate band; studies that break into High typically combine multi-center datasets, prospective evaluation, and open-source release.

§06Full Metadata
Citation & identification
Bibliographic record
▸
Model IDoculoplastics_chen_2021
Full author listTzeng, S. S., Hsiao, Y. C., Chen, R. F., Hung, E. C., & Lee, O. K.
JournalJMIR Mhealth Uhealth
DOI10.2196/32444
ConditionsBlepharoptosis
Architecture
Model internals
▸
Architecture familyCNN
Architecture detailEnsembled CNNs within the MAIA platform, preceded by MobileNetV2 reflex localization.
Funding & conflicts
Disclosures
▸
FundingNR
Conflicts of interestNo conflicts
Card lifecycle
Status, version, reviewers
▸
Card statusPublished
Card versionv1.0.0
First added2026-04-28
Last reviewed2026-04-28
Reviewersattribution opt-in pending
Update history
Card changelog
▸
  • 2026-04-28 initial publication (v1.0.0)
← Back to Oculoplastics