**4. Discussion**

Chatbots have undergone a terrific evolution through decades of technological innovation. The latest and most impactful advancement is the one we are experiencing, which is represented by artificial intelligence [8,9]. Today, the chatbots can find up to 37 applications [33]. Population expectations of chatbots are many and important. The major expectation of the population with respect to these systems is chatbots' rapid response in emergency situations [33]. The greatest expected benefit is that of being able to receive 24 h service from these systems. The main "expectations" and the main "expected benefits" of different domains seem to overlap, including the very important one of healthcare. In fact, specifically in healthcare, having a rapid response in the event of an emergency, in a 24 h mode, is strategic in emergency medicine and intensive hospitalization medicine. Additionally, *healthcare* is one of the domains affected by the introduction of these systems. After having retraced, through our study, the evolution of technology, and after having addressed the topic of applications, from which expectations of use and cross-benefits between categories (including the health domain) have emerged, we have focused on the *health domain*.

An overview on PubMed confirmed a rapid increase in scientific interest in the health domain [68], in line with the perceived general interest [69]. In particular, PubMed showed that in the last three years, a volume of publications equal to 91.2% of total publications on these systems has been published, which, as has been discussed, have existed since 1966 [13]. A substantial contribution to this growth was also made both by the COVID-19 pandemic, which has brought 91 publications since 2020 [70] (see the composite key in the first position of the Box 1 reported below), and by artificial intelligence, which was of interest in 123 studies [71] (see the composite key in the second position of the Box 1 reported below).

The topics of greatest interest emerged from a search on PubMed based on systematic reviews. These studies concerned two specific systematic reviews on *technology assessment* and systematic reviews analyzing chatbot applications in *mental health, addiction, chronic diseases, wellness and fitness, and, finally, the applications on heterogeneous applications in the health domain and technology assessment.*

The two studies [54,59] on *technology assessment*, even if limited to a few domains, gave us important indications on chatbot technology.

The first one [59] highlighted the lack of a standard procedure to study usability, with a variety of different tools and metrics used to assess usability with little consistency in the study designs. The authors recommend the development of a standardized procedure for evaluating usability.

The second [54], while acknowledging the potential effectiveness of these systems, deemed it necessary to further investigate the real requirements that make chatbots effective and the potential benefits.

The other studies have unanimously highlighted the potential opportunities of these systems in specific applications but have highlighted various *critical issues concerning different single domains.*

The need for more attention on the *domain of ethics* has been recommended, for example in [56], with regard to the applications in healthcare education, research and practice.

The need to deepen the *domain of safety*, to avoid the potentially harmful impact of these systems has been highlighted in [62], with a focus on the domain of drug addiction.

The need for more attention on the *domain of standardization* has been highlighted by three studies [40,49,58]. In [58], the development of specific guidelines was recommended in a *chronicity study*. Reference [49] suggested the standardization of procedures and protocols in the domain of *addiction*. Even in *applications dedicated to fitness and wellness*, the need for standardization has been recalled both in design and reporting [40].

Concerns regarding the *domain of efficacy* have been raised in several studies [40,45,51,57,60,61,64,65,67]. An in-depth study of this domain has been suggested in various applications of mental health [45,61,65,67] in the domain of addiction [57], in the field of chronicity [60], in post-intervention medicine [51] and in wellness and fitness [40,64].

The *domain of cybersecurity* has been touched upon to a certain extent by all the studies, but particularly in two studies that have addressed multiple applications simultaneously in the health domain [52,56]. The *domain of interdisciplinarity* was recalled as important in cancer therapy, where this aspect, as is known, is a key factor [63].

The *need to address some domains* together was highlighted in [56]. In this study, dedicated to AI-based chatbots used in healthcare education, research, and practice, concerns were expressed on *ethical aspects, copyright, transparency, legal issues, the risk of bias, plagiarism, lack of originality, cybersecurity issues and inaccurate content*.'

**Box 1.** The proposed composite keys.

*("chatbot"[Title/Abstract] AND ("health"[Title/Abstract] OR "healthcare"[Title/Abstract] OR "health domain"[Title/Abstract]) AND ("COVID-19"[All Fields] OR "COVID-19"[MeSH Terms] OR "COVID-19 vaccines"[All Fields] OR "COVID-19 vac-cines"[MeSH Terms] OR "COVID-19 serotherapy"[All Fields] OR "COVID-19 nucleic acid test-ing"[All Fields] OR "COVID-19 nucleic acid testing"[MeSH Terms] OR "COVID-19 serological testing"[All Fields] OR "COVID-19 serological testing"[MeSH Terms] OR "COVID-19 test-ing"[All Fields] OR "COVID-19 testing"[MeSH Terms] OR "SARS-CoV-2"[All Fields] OR "SARS-CoV-2"[MeSH Terms] OR "severe acute respiratory syndrome coronavirus 2"[All Fields] OR "ncov"[All Fields] OR "2019 ncov"[All Fields] OR (("coronavirus"[MeSH Terms] OR "coro-navirus"[All Fields] OR "cov"[All Fields]) AND 2019/11/01:3000/12/31[Date–Publica-tion]))) AND (2020/1/1:2023/4/14[pdat]) ("chatbot"[Title/Abstract] AND ("health"[Title/Abstract] OR "healthcare"[Title/Abstract] OR "health*

*domain"[Title/Abstract]) AND ("artificial intelligence"[MeSH Terms] OR ("artificial"[All Fields] AND "intelligence"[All Fields]) OR "artificial intelligence"[All Fields])) AND (2020/1/1:2023/4/13[pdat])*
