Diagnostic model, 2,140 images

Machine learning research poster

A model that detects diabetic retinopathy from smartphone fundus images, laid out as an A0 landscape board. It is built around the figure, which is what a machine learning poster usually needs, and the text does the two things these posters most often skip: it describes the split honestly, and it says plainly what the result does not show.

Create a poster with OneCraftA0 Landscape, printed at 1189 by 841 millimetres

The whole board

The poster at full size, exactly as it prints. Every number, citation and caption on it was written for this example, so the layout is being judged on real content.

Machine learning detection of diabetic retinopathy from smartphone fundus images in low resource clinics

J. Chen ¹, M. Garcia ¹ ², T. Adebayo ²

1 Department of Computer Science, Northgate University · 2 Institute for Tropical Medicine, New Kent

ABSTRACT

Referral for diabetic retinopathy depends on a trained grader and a tabletop fundus camera, neither of which is present in most district clinics. Patients are therefore referred on symptoms, by which point the disease that is easiest to treat has usually been missed. We trained a convolutional model on 2,140 images captured with a $60 smartphone lens adapter and graded by two ophthalmologists, and tested it on a held out set of 428 images from three clinics that contributed no training data at all. Sensitivity for referable disease was 91.2% (95% CI 86.2 to 95.1) and specificity 88.7% (95% CI 84.0 to 92.4), measured against a two grader reference standard with a third grader adjudicating disagreements.

METHODS

Data. 2,140 images from 1,070 patients, both eyes, captured on a smartphone with a clip-on 20 dioptre lens.

Reference. Two ophthalmologists graded independently on the international scale; a third adjudicated disagreements.

Model. EfficientNet-B0 pretrained on ImageNet, fine tuned for 40 epochs, images resized to 512 by 512.

Split. Patient level split. The test set came from three clinics that contributed nothing to training.

Analysis. Sensitivity and specificity for referable disease, with bootstrap confidence intervals over 2,000 resamples.

91.2%

sensitivity for referable disease

95% CI 86.2 to 95.1, n=171 referable eyes

Sensitivity 91.2%, specificity 88.7% on unseen clinics

Runs on a $60 lens adapter and an existing phone

Errors driven by media opacity, not disease grade

Quality filter lifts sensitivity to 94.8%

Sensitivity

Specificity

Figure 1. Sensitivity and specificity by disease grade on the held out test set.

Source: Trial dataset, 2026.

RESULTS

On 428 held out images the model reached 91.2% sensitivity (95% CI 86.2 to 95.1) and 88.7% specificity (95% CI 84.0 to 92.4) for referable disease. Performance was weakest on mild disease, where sensitivity fell to 78.6%, and strongest on severe and proliferative grades, where it exceeded 95%. That pattern matters less than it first looks. Mild disease is not referable under the local pathway, so the misses that carry clinical cost are concentrated in exactly the grades the model handles best, and the mild cases it misses are returned to routine annual screening rather than lost. The test set held 171 referable eyes and 257 that were not referable, so 91.2% sensitivity means 15 misses. Of those 15, nine came from images the graders had already flagged for media opacity, mostly cataract, and three from images where the flash reflected off the lens adapter. Only three were failures on a clear image. Excluding images that failed an automated quality check lifted sensitivity to 94.8% while discarding 11.2% of the set. That is the trade a clinic has to weigh: a better model on fewer images, against the cost of asking one patient in nine to come back. In a district where the return journey is an hour each way, the arithmetic may favour accepting the lower figure. Inter grader disagreement on the same 428 images was 7.9%. The model is therefore wrong about as often as the two ophthalmologists were wrong about each other, which sets a practical ceiling on what any further tuning against this reference standard can achieve. Throughput was 1.8 seconds per image on a mid range phone with no network connection, so the pathway does not depend on a clinic having reliable data, and no image leaves the device. Two things this study does not show are worth stating plainly. The first is anything about outcomes. We measured agreement with a grading standard, not whether a single patient was treated earlier or kept their sight, and the step from one to the other runs through a referral pathway we did not test. The second is generalisation beyond this population.

Block by block

What each block on the board is for, in the order a reader walks it.

Title band and authors: Title, generated author superscripts and both affiliations, on a single band across the top of the landscape board.
Abstract: The referral problem in district clinics, the 2,140 smartphone images graded by two ophthalmologists, the held out set of 428 images from three clinics that contributed no training data, and the headline sensitivity and specificity with their intervals against an adjudicated two grader reference.
Methods as a labelled protocol: Five entries: data, reference standard, model, split and analysis. The split entry is the one reviewers stop at, so it says explicitly that the test clinics contributed no training data.
The dominant figure: Sensitivity and specificity by disease grade across the full width of the board, which is the figure a reader will look at before reading anything.
Key number: One stat block carrying 91.2% sensitivity with its interval, rather than a row of metrics competing for attention. The line under it names the sensitivity denominator, 171 referable eyes rather than the 428 image test set, and the percentage follows from it: 15 misses leave 156 of 171, which is 91.2%.
Results: One long results block: the headline metrics, the weakest grade at 78.6% sensitivity for mild disease, the breakdown of the 15 misses, what the quality filter trades, the 7.9% disagreement between graders as a ceiling, and throughput of 1.8 seconds per image on a phone. The weakest grade is reported rather than buried.
What it does not show: The results block ends on the two limits: no outcome was measured, only agreement with a grading standard, and the result is not shown to generalise beyond this population. The second limit is announced in a single sentence; on your board, say which populations, devices and sites the test set did not cover.
Takeaways: Four checked points: the two headline metrics on unseen clinics, the $60 lens adapter on an existing phone, errors driven by media opacity rather than disease grade, and the quality filter lifting sensitivity to 94.8%. This layout carries no reference block, so cite any sources inside the methods or results text.
How to adapt this board: Describe the split before anything else: how many cases, from where, and what the model never saw. On a diagnostic board that sentence decides whether the rest is worth reading. Then pick the one curve that is the result and give it the hero figure, because a board with four curves has no result. Errors go by class in the results text, since an overall accuracy hides the failure that matters clinically. The key number block takes the single headline metric with its interval and the denominator it was computed on.

What makes this board work

The split is described, not assumed

The methods block states that the split was at patient level and that the test set came from three clinics contributing nothing to training. Most model posters say "held out set" and leave the reader to guess whether the same patient appears on both sides.

Errors are explained, not just counted

Fifteen misses are broken down into nine with media opacity, three with flash reflection and three genuine failures on a clear image. That turns a number into a direction for the next piece of work.

It states what it does not show

The results block says outright that no outcome was measured and that the findings are not shown to hold beyond this population. A poster that names its own limits is more persuasive than one that does not, because reviewers ask anyway.

Questions people ask

What goes on a machine learning research poster?

The task and why it matters, the dataset with its size and how it was labelled, the split, the model, the metrics with confidence intervals, an error analysis, and the limits. A confusion matrix or a curve usually carries the result better than a table.

Should I show accuracy on a poster?

Rarely on its own. For a screening task, sensitivity and specificity with intervals say what accuracy hides, because a model can score well on accuracy by rarely predicting the rare class. This test set held 171 referable and 257 non referable eyes, so a model that called every eye healthy would still look respectable on accuracy while missing every patient.

Landscape or portrait for a machine learning poster?

Landscape suits a single dominant figure and a wide reading order, which is the usual shape for a model result. This board runs the abstract and methods on the left, the stat block and figure across the middle, and one long results block on the right. Portrait suits a text heavier study. Check what the conference specifies before choosing either.

Can I put my own figure on it instead of a chart?

Not on this layout. The hero figure is a chart block, so it takes one of ten chart kinds with your own values rather than an image file. Four layouts carry image figures, but those slots cannot be filled yet either, so plan on charts and paste your numbers.

Which chart kind suits a ROC or PR curve?

Line, with one series per model and the operating point named in the caption. The chart block accepts ten kinds, and line is the only one that keeps a curve honest. Never put a curve in an area chart, which fills under it and hides the comparison.

Should I list hyperparameters?

Only the ones that matter for reproduction, in one methods entry. This board names the architecture, EfficientNet-B0 pretrained on ImageNet, 40 epochs of fine tuning and a 512 by 512 input size. Learning rate, batch size and seed would complete that entry. Everything else belongs in a repository or supplement, because a board that lists forty settings is a board nobody reads.

Where does the data statement go?

Near the limits, as a sentence that says what the model did not see: sites, devices, years or populations missing from training. This board has no discussion block, so its results text carries it: three test clinics that contributed no training data, one smartphone lens adapter, and a closing line that the result is not shown to generalise beyond this population.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make one like this

Sources

Written and checked by the OneCraft team. Last checked 29 September 2026.