Lig-Doctor: Real-World Clinical Prognosis using a Bi-Directional Neural Network

Automated medical prognosis has gained interest as artificial intelligence evolves and the potential for computer-aided medicine becomes evident. Nevertheless, it is challenging to design an effective system that, given a patient's medical history, can predict probable future conditions. Previous works have tackled the problem by using artificial neural network architectures that do not benefit from bi-directional temporal processing, or by utilizing non-generalizable inference approaches. Differently, we introduce a Deep Learning architecture whose design results from an intensive experimental process; our final architecture is based on two parallel Minimal Gated Recurrent Unit networks working in bi-directional manner, which was extensively tested with two real-world datasets. Our results demonstrate significant improvements in automated medical prognosis, as measured with metrics Precision@, Recall@, F1-Score, and AUC-ROC. We contribute with an architecture and with insights for the design of Deep Learning architectures.