The challenge of deciphering doctors\' handwriting on prescriptions has long been a persistent issue in the healthcare field. Illegible handwriting can lead to medication errors, incorrect dosages, and delays in treatment, all of which may compromise patient safety and overall healthcare outcomes. Finding a reliable solution to this problem is therefore vital to improving both efficiency and accuracy in medical services.
Previous studies have examined the legibility of doctors\' handwriting in various languages and employed machine-learning techniques to recognize handwritten text. However, many of these approaches remained theoretical, with limited application in real-world pharmacy settings where checking medicine availability and streamlining transactions are essential.
This research proposes a kiosk-based system that leverages deep learning to automatically recognize handwritten medicine names from prescriptions. Once identified, the system cross- checks the medicine against a pharmacy database to confirm availability and generates an order receipt that can be presented at the pharmacy counter. This process reduces delays and minimizes the risks associated with manual interpretation.
The prototype system is powered by a Raspberry Pi using Python, OpenCV, and TensorFlow. Trained with a dataset of 1,400 handwritten medicine names and the IAM dataset, it achieved a recognition accuracy of 91.66%. By automating prescription reading and medicine availability checking, the kiosk enhances accuracy, speeds up service, and reduces medication errors, thereby promoting safer and more efficient patient care.