Implementation of Chatbot that Predicts an Illness Dynamically using Machine Learning Techniques

Abstract

Timely access to healthcare is crucial in order to maintain a high standard of living. However, obtaining medical consultations can be difficult, especially for those living in remote areas or during a pandemic when face-to-face consultations are not always possible. The ability to accurately diagnose diseases is essential for effective treatment, and recent technological advancements offer a potential solution. Machine learning (ML) and Natural language processing (NLP) enables computer programs to understand human language and extract desired features from responses, allowing for human-like interaction with users. By leveraging these technologies, healthcare professionals can potentially provide more accessible and efficient medical consultations to individuals, regardless of their location. The concept is to establish an online platform where users can ask medical-related queries and receive responses from both medical professionals and fellow users. The platform would feature a Medical Chatbot, which employs advanced ML techniques to analyze user-provided symptoms and provide initial disease diagnosis and related information prior to consulting with a doctor. This disease prediction chatbot interacts dynamically with the users to enter the symptoms of the diseases and based on syntactic and semantic similarity response is given. In this work the threshold of similarity score is kept of 0.7. K-Nearest neighbors, Random forest, Support vector machine, Naive bayes and Logistic regression algorithms are used for prediction of disease based on symptoms which are faced by users. The syntactic similarity, fuzzy string matching and semantic similarity using all-MiniLM-L6-v2 model is used to improve the efficiency of the result.

Digital Assistant for Pharmacists Using Indonesian Language Based on Rules and Artificial Intelligence

“…However, creating chatbots capable of handling sensitive information, particularly in healthcare, presents unique challenges. Errors made by these bots when extracting information or providing health advice can lead to significant consequences (2). In developing a chatbot for pharmacy recommendations, it becomes crucial for the bot to proficiently extract symptom-related information and other pertinent patient data proficiently, subsequently offering tailored recommendations or medication suggestions based on this data.…”

“…In this study, we proposed the usage of regular expression matching (3,4) to create an Indonesian chatbot for pharmacy recommendation, SmartFarma. Prior research shows that the regular expression matching method is more transparent and accurate than other machine learning algorithms (2,(5)(6)(7). This choice is rooted in the recognition that, especially within the healthcare domain, transparency in model results is of paramount importance (3,4,8).…”