Loading model and data…

// ML · Browser inference (samples)

Pediatric chest X-ray: NORMAL vs PNEUMONIA

Educational portfolio demo: a small MobileNetV2-based classifier runs entirely in your browser on curated public-domain samples. Not for clinical use.

Non-medical disclaimer This page is for demonstrating machine learning on a public dataset only. It is not a medical device, not FDA cleared, and must not be used for diagnosis, screening, or treatment decisions.

Sample image

Selected chest X-ray sample

Gallery

Model output

Prediction

—

Vs demo sample label

—

Confidence (predicted class)

—

Limitations (this sample)
Model card (lite)

Input: — · Preprocessing: —

Backbone: — · Split: —

Held-out test metrics

MetricValue

Confusion matrix (test)

Pred NORMALPred PNEUMONIA

Limitations

    // Technical summary

    What this demo uses

    TensorFlow.js JavaScript Browser inference CNN

    This demo runs a small image-classification model in the browser on sample chest X-rays to show how medical AI interfaces can be presented responsibly.

    • Methodology: A pre-trained binary model is converted for TensorFlow.js and loaded locally. Inference is run only on bundled sample images, with transparent test metrics shown in the model card.
    • Technical terms:
      • CNN: Convolutional Neural Network, a model type for images.
      • Inference: The model making a prediction.
      • Confusion matrix: A table showing correct vs incorrect predictions by class.
    • Toolsets/technology used: TensorFlow.js, JavaScript, browser-based model inference, and static sample assets.