Conception and Realization of a Chatbot-System to support Psychological and Medical Procedures

Universität Ulm

MA Abschlussvortrag, Jens Winkler, Ort: O27/5202, Datum: 18.02.2019, Zeit: 14:00 Uhr

As a result of long term researches of Artificial Intelligent, the influence as well as the amount of possible use-cases of chatbots is growing constantly. As to health care, chatbots can be used to simplify the interaction between experts and patients. For example, chatbots can help people who are affected by depression, as the barrier to chat with an application is less high than meeting a real person. While time flexibility can be a benefit for patients, the response to frequently asked questions relieves experts so they can focus on important issues. Consequently, the usage of chatbots in psychological and medical sectors can be a step to improve the health care system. The objective of this master’s thesis is to create a system concept involving a conversational agent, a mobile application and a back-end application to built solutions for the psychological and medical sectors. As all possible use-cases can not be clearly restricted, the concept is designed to be adjustable. Referring to health care, the systems needs to be able to react to critical situations like medical emergencies. However, handling life-threatening cases exceeds the capability of the system and always needs human assistance. Therefore, the concept also includes human participants such as experts. To test the approach, a mobile application was developed that includes a simple conversation using the AI assistant IBM Watson Assistant. In addition, a generic framework was developed to handle multiple third-party chatbots. Furthermore, the mobile application reacts to critical situations, such as the detection of suicidal thoughts or if the user enters an input that is unknown for the system. Moreover, the knowledge base of the chatbot can be extended using a conversation formatted process.