Digestive Symptom Classification App

This app allows users to input their symptoms through a guided questionnaire. The inputs are then processed by an LSTM model (compressed to TensorFlow Lite) on-device to classify the likelihood of various digestive conditions and provide health guidance.

The LSTM model was trained in Python/Keras and converted to TFLite for efficient on-device inference within the native Android app.

I built both the machine learning pipeline (data preprocessing, model training, and TFLite conversion) and the full Android application including UI, on-device inference, and local data storage.

• Symptom input via guided questionnaire
• On-device LSTM inference with TFLite
• Multi-class digestive disease classification
• Health recommendation and guidance output
• Offline functionality — no internet required