Automatic medical report generation and specialist referral prediction

The integration of artificial intelligence in clinical workflows can reduce medical workload and improve early detection of conditions that require specialized care. This research presents a system for automated medical reporting and referral recommendation. It leverages synthetic electronic health record (EHR) data generated with Synthea. The system can predict the presence of multiple diseases based on clinical observations and determines if a patient should be referred to a specialist. Two machine learning approaches are evaluated: a traditional Random Forest classifier and a Single Layer Feedforward Network (SLFN). The predicted conditions are then used to drive the decision to refer a patient. The output of this process is used to automatically generate a report using ChatGPT. Experimental results show that both models achieve competitive performance on common clinical conditions. This approach demonstrates a scalable framework for testing diagnostic and referral systems using synthetic data. Keywords: Artificial intelligence; Random forest; SLFN; Synthetic EHR data; Automated medical reporting; ChatGPT; Referral recommendation;