Predicting Consultation Success in Online Health Platforms Using Dynamic Knowledge Graphs and Multimodal Data Fusion

In virtual or online health platforms, accurately predicting the success of online consultations is paramount in the face of fierce competition. The scarcity of patient data poses a significant challenge to predicting online healthcare consultation success rate. To address this challenge, we introduce MDKSP, which harnesses advanced language models, a novel Knowledge Graph Attention Network, and a new multi-modal data fusion technique. MDKSP enhances predictive accuracy by capturing explicit (patient-doctor communication) and implicit (digital traces in patients' healthcare journeys, both online and offline) knowledge. MDKSP significantly enhances the predictive capability of healthcare consultation success in virtual health. MDKSP's utility extends to diverse virtual or hybrid models, such as online education (predicting student retention at the onset of a course), hybrid sales (forecasting purchase intent through online information provision and offline testing and real-world experiences), and more.