ChatHans: sistema de pré-diagnóstico da hanseníase

Abstract

Leprosy is a chronic infectious disease that, when not diagnosed early, can lead to permanent physical disabilities. In Brazil, early diagnosis still represents a challenge, especially in areas that are difficult to access and have limited health infrastructure. In this context, this research proposes the development of ChatHans, a hybrid Chatbot to assist in the pre-diagnosis and suspicion of Leprosy, using the Leprosy Suspicion Questionnaire (QSH), tools based on artificial intelligence and natural language processing. The methodological approach adopted follows Design Science Research (DSR), combining machine learning techniques, computational modeling, and qualitative analysis with health professionals. Initially, experts were interviewed to gather requirements based on the QSH and the guidelines of the Brazilian Ministry of Health. From this data, a hybrid Chatbot was structured with three main functionalities: (i) "Assess Leprosy Risk", applying the QSH to estimate the suspicion of the disease; (ii) "Clearing Doubts", providing information about the disease and its treatment; and (iii) "Testing Skin Sensitivity", helping users perform basic tests of thermal and pain sensitivity. The system was implemented using the Blip platform, integrating with DialogFlow for natural language understanding, with a Web interface (Chat-Blip) and Telegram. A scoring model was developed to assign risk levels based on users’ responses to the QSH, reinforcing suspected cases for referral to health services. The results demonstrate that the Chatbot can be an interesting tool in active leprosy screening, automating such a process. Usability tests indicated a high acceptance rate, in which the average score of the experience with the proposed system was 8.6 (between 0 and 10), with a low standard deviation of 1.04, with the sample scope accounting for 92 participants. However, challenges related to the complexity of some interactions were also identified. Furthermore, the analysis of the Chatbot interactions showed that most users can complete the risk assessment without difficulty, indicating the system’s viability for practical application. The originality of this research lies in applying hybrid Chatbots in the screening of Leprosy, one of the neglected diseases that affect millions of people worldwide, combining rule-based approaches and artificial intelligence to optimize the pre-diagnosis process and clarify doubts about the pathology. Unlike other tools, such as applications to support clinical diagnosis, ChatHans seeks to actively engage the user in the self-assessment of symptoms and facilitate referral to the health system. This research also offers a new scoring metric for the Leprosy suspicion questionnaire that considers statistical criteria for confirming cases presented in other studies, indicating three levels based on percentiles and sample variance of the studies that apply this instrument.