A Acupuncture has become a popular alternative medicine method in Indonesia, especially for patients avoiding surgery and seeking a \\\"body lightening\\\" effect. This research aims to develop an early detection system for cardiovascular and gastrointestinal disorders using photoplethysmography (PPG) sensors on acupuncture points to help practitioners predict patients\\\' initial conditions. Three PPG sensors were placed on Taiyuan (LU9), Jingqu (LU8), and Shenmen (H7) points, recording 60-second pulse data from 369 respondents. A Convolutional Recurrent Neural Network (CRNN) model was developed using TensorFlow to classify pulse patterns. Results showed 95% training accuracy uisng train-test split validation, but 60-73% validation accuracy using k-cross validation, indicating overfitting. After conversion to TensorFlow Lite, model accuracy became 78% due to quantization. Testing on 281 unverified samples showed a tendency for the model to predict the majority of samples as gastrointestinal (75.44%) and cardiovascular (21.00%) disorders. Testing on 5 subjects outside the measurement data demonstrated promising performance, although with some inaccuracies. The system was integrated into a Flutter application for ease of use. This research opens opportunities for developing non-invasive diagnostic tools based on Traditional Chinese Medicine, but requires increasing the dataset size, optimizing the model to reduce overfitting, and further clinical trials to improve accuracy and reliability in acupuncture practice.