NeoJaundice-AI detects newborn jaundice via smartphone in 3 seconds offline
91.8% accuracy, 93.5% sensitivity – no blood test needed for neonatal jaundice screening.
NeoJaundice-AI is a smartphone-based screening system for neonatal jaundice (hyperbilirubinemia), a condition affecting ~15 million newborns annually in India alone. Developed by researchers Rahul Patel and Nirjala Jarpula, the system uses a dual-branch EfficientNet-B0 architecture that independently processes photos of the baby’s skin and sclera (eye whites). Deep features are fused with handcrafted YCbCr color statistics to perform both four-class severity classification and continuous bilirubin regression. A key innovation is a synthetic jaundice generation method that simulates bilirubin-induced yellowing via controlled YCbCr channel modifications on normal neonatal skin images, specifically addressing data scarcity for severe cases and darker Indian skin tones (Fitzpatrick Types IV to VI).
Experimental results show an overall classification accuracy of 91.8%, clinical sensitivity of 93.5%, and a bilirubin mean absolute error of 1.4 mg/dL. After INT8 quantization and ONNX conversion, the model size is reduced to 8.3 MB, maintaining inference times below three seconds on standard Android devices without internet connectivity. This is the first India-focused neonatal jaundice AI system combining multimodal image fusion, skin-tone adaptation, synthetic data augmentation, and fully offline mobile deployment in a single framework. The system has the potential to dramatically improve early detection in rural clinics where blood tests are impractical.
- Achieves 91.8% classification accuracy and 93.5% sensitivity for jaundice severity using only smartphone photos.
- Model runs fully offline on Android devices with inference <3 seconds and a compact size of 8.3 MB.
- Synthetic augmentation and skin-tone normalization ensure reliable performance on darker Indian skin tones (Fitzpatrick IV–VI).
Why It Matters
Enables newborn jaundice screening in low-resource rural clinics without lab access, potentially saving thousands of lives.