To set the scene with a quadruple Anglicism, chatbot, deeplearning, computer vision and extend reality have radically become the watchwords of our century. Our thesis aims to make a contribution in the field of medical conversational agents which has become a key factor in self-medication, in the field of biometric authentication by the iris (one of the most precise and difficult to hack modalities), face (one of the least intrusive and least expensive modalities) and that of ensuring appropriate management of stroke patients. This involves immediate measures upon admission of the patient, revascularization procedures, initiation of prevention and analysis of acute complications, assessment of etiological factors and initiation of rehabilitation while considering the patient in its entirety. Through this work, we address several important aspects of mono and multimodal biometrics. We begin by drawing up a state of the art on monomodal biometrics by the iris and by the face and on the multimodality iris/face, before proposing several personal approaches to individual recognition for each of the two modalities. We approach, in particular, facial recognition by classical approaches based on combinations of algorithms and bio-inspired approaches emulating the mechanism of human vision. We demonstrate the interest of bio-inspired approaches compared to classical approaches through two methods. The first exploits the results from neuroscientific work indicating the importance of regions and decomposition scales useful for identifying a face. The second is to apply a rank-order coding method in the pre-processing phase to enhance the information content of face images. With regard to the management of patients who have suffered a stroke, we will broaden our fields of study to address more prevention and analysis of acute complications, finally to determine the risk factors and thus consider taking in early care. Processing software, diagnostic algorithms, user interfaces, and systems for collecting and disseminating information between all the actors concerned. The proposed devices are designed to be the least invasive and intrusive possible. For better interaction with the patient, a chatbot will accompany our interfaces as a conversational agent, since in the fields of artificial intelligence (AI) and big data, the latter make it possible to collect and use as much data as possible to offer a better service. efficient and more personalized. We also praise their better accessibility compared to existing tools, because they allow us to access a service without having to open a mobile application or an Internet browser. Finally, researchers have observed that individuals are willing to talk to robot assistants longer than to human beings, and also more tempted to confide their intimate secrets to them.
Chatbot, deeplearning, computer vision, biometric authentication, stroke, algorithm, facial recognition, bio-inspire, neuroscientist, artificial intelligence, bigdata, robot-assistant.