Many of our day-to-day interactions are now facilitated by voice-activated assistants like Google Assistant and Amazon Alexa. In addition to these all-in-one solutions, there is an increasing need for the development of conversational systems in healthcare. Such systems have a common design concept: a personal secretary who offers healthcare via natural dialogue. The primary goal is to develop an agent that guides a patient via a turn-taking conversation just as doctors do with patients. With chatbots, this paradigm change is taking place as an inexpensive way to supply these services.
Chatbots are automated systems that mimic the behavior of users on one side of a chat conversation as emulation systems that duplicate the sound and feel of two people conversing in real-time. They serve as a model for communicating with the other end of the line efficiently and intelligently. In addition, they help cultivate healthy self-care practices that contribute to a sense of well-being. Because of the conversation model, chatbots are easy to use and can be accessed from any smartphone, anywhere and at any time. Nevertheless, the provision of healthcare still relies on the expertise of medical experts.
A chatbot for healthcare can be envisioned as a series of nested layers. The user and domain databases are stored within the knowledge layer from which information is fed to the service layer of healthcare delivery, which implements mechanisms for clinical decisions. Next, the dialogue layer receives this information as soon as the decisions have been generated. The dialogues prepared using rule-matching methods are resilient and simple to construct, but are limited to a narrow domain; probabilistic (machine learning) techniques may produce more realistic dialogue but are less robust. The dialogue layer gathers user intents, generates replies by contacting the service layer, and transmits them to the presentation layer, which provides a text- or voice-based user interface.
The commercially available chatbots mainly work in three domains in healthcare: diagnosis, prevention and treatment of disease.
Chatbots for diagnosis support
Chatbots for diagnosis assess the user's symptoms and propose treatments. Three broad types of diagnostic chatbots include those providing diagnostic support, chatbots checking specific symptoms and systems checking general symptoms. The first type does not perform diagnostics but rather supports a diagnosis through online consultations, making available health services, or by offering conversational access to data on diseases and symptoms. A specific symptom checker, the second type, seeks to either assist users in confirming the severity and presence of an illness or diagnose a specific disorder, similar to a discussion with a doctor. The last type, the general symptom checker, imitates a discussion with a general health expert by guiding patients via a set of questions about their symptoms to identify a problem and, in certain circumstances, recommending a line of treatment.
Chatbots for preventing diseases
The chatbots in this capacity aid in monitoring and educating patient about health, as well as preventing problems by promoting healthy behaviors. The three major types of chatbots in this category are those providing access to healthcare services, offering health education and providing coaching. Chatbots providing access to healthcare do not engage in the delivery of healthcare services but rather serve as a point of access for these services. This is done by connecting people with healthcare providers, locating medications online, and performing healthcare customer service duties. The educational chatbot assists in preventing diseases by teaching consumers how to prevent certain health concerns. The last chatbot type, which provides coaching, aims to avoid health deterioration by fostering an active lifestyle and enhancing overall well-being.
Chatbots for treatment
These chatbots aid or offer treatment for particular health disorders (such as pregnancy or therapeutic diet). Similar to the above chatbot types, these can also be sub-categorized into systems providing treatment support, health treatment and cognitive-behavioral treatment. The first chatbot helps during the treatment stages; examples include individualized medication adherence alerts as part of treatment and a list of drugs based on good reviews from users on the internet for natural health remedies. The therapist chatbot assumes more prominent participation by offering in-home therapy to patients. They give either practice-based or drug-based treatment to their patients, depending on their major focus. The last one, the cognitive behavioral therapy chatbot, encompasses a spectrum of treatments that target particular emotions and mental states. The treatment is a guided, organized dialogue that begins with a question-and-answer session to diagnose the patient's issue. It then suggests appropriate exercises depending on the estimated conditions and monitors the intended state.
Existing health chatbots continue to serve as a supplement rather than a substitute for medical personnel. The many forms of chatbots discussed in this article play a continuous role in healthcare provision by simulating activities of a doctor, easing entry to quality health care by connecting people with hospitals or assisting delivery of services, and supplying patients with goods and information.