ABSTRACT
Healthcare is a sector that is expeditiously developing in technology and services. In recent years, the Covid-19 pandemic has drastically affected the working of the healthcare sector; people are apprehended to visit hospitals for any treatment. But evolution in modern technologies has opened multiple paths to improve and modernize the working of the healthcare sector. The proposed system is a multi-layered disease prediction model that analyzes numerous factors for predicting diseases. The system analyzes the symptoms using a modified decision tree algorithm that predicts the possible illness and suggests the test accordingly. The model is trained individually for each type of test format. For image type, reports were classified with convolutional neural networks. For PDF type, the data was extracted using optical character recognition (OCR). The model uses the Levenshtein distance to find unigrams and bigrams. The match is further analyzed, and a detailed summary of the report gets generated. Report summary and the predicted disease are provided to the patient with the list of home remedies. Further, a specialized doctor receives all the medical diagnosis details when a patient books an appointment. Hospitals usually face the problem of patient versus nurse ratio. It creates management issues to the critical ward. Patients are left unattended and can cause death threats. The proposed system analyzes multiple and dynamic factors. It increases the accuracy of the prediction. The proposed 36hospital monitoring system observes the vital signs on the patient monitors beside the ICU beds and notifies the hospital staff after encountering the abnormality. The model dynamically calculates the threshold value for each vital sign considering multiple factors like age, gender, and medical history of the patient. By understanding the patient’s current medical condition, the model responds to change in vital signs and gives an idea about the organ’s condition. Machine learning algorithm – random forest regression helps in calculating the threshold values of heart rate (HR) and respiratory rate (RR). Equations for blood pressure (BP) get the threshold values depending on age and gender. These custom thresholds for specific patients reduce false alarms, which was a significant concern in the previous monitoring system.
