Study 1

Studiul 1, efectuat pe datele colectate de la 234 de supravie?uitori de cancer, a ar?tat c? gdurile intruzive sunt un nod central al re?elei fric? de recuren?? – depresie – anxietate. Dintre toate formele de fric? de recidiv?, gândurile intruzive exercit? cea mai mare influen?? asupra re?elei mai sus men?ionate. Acest rezultat impacteaz? studiile viitoare ?i implica?iile lor practice prin faptul c? orienteaz? evalu?rile ?i interven?iile viitoare punctual înspre centrarea pe gândurile intruzive ca mecanism al schimb?rii fricii de recuren??. Acest rezultat poate da form? interven?iilor viitoare asupra fricii de recuren??, interven?ii care la ora actual? au un efect modest, chiar mic asupra remedierii fricii de recuren?? a cancerului. 

Another interesting result that emerged from Study 1 is related to the comorbidity of fear of recurrence with depression. Dysfunctional coping style, another dimension of fear of recurrence, is linked with depression only in already vulnerable cancer subgroups. By this we mean short-term survivors diagnosed under the age of 18 or survivors in advanced stages of cancer or those still undergoing treatment. For them, focusing on functional coping strategies is likely to be more beneficial than for less vulnerable groups. Thus, the study results inform tailored interventions for cancer survivors in terms of the core forms of fear of recurrence for different risk groups. They also remind us that there is no one-size-fits-all solution for assessing and intervening with cancer survivors, rather, tailored solutions are recommended. 

Study 2

Study 2, a meta-analysis of 72 primary studies, has important implications for cancer survivors, healthcare providers and researchers through its results. First, the moderately positive relationship between fear of cancer recurrence (FRC) and mental health confirms that FRC is a prevalent problem among cancer survivors, which is associated with a wide range of problems across the spectrum of anxiety, depression and emotional distress. The results indicate that FRC should be included in protocols to assess the mental health of cancer survivors and vice versa. They also raise awareness of the psychological and emotional impact of a cancer diagnosis. Last but not least, the meta-analysis draws attention to the fact that only a third of the studies included in it focused primarily on mental health, in most cases this was measured secondary to other objectives, i.e. none of the studies had a clinically diagnosed group. Therefore, assessing mental health as a priority for the survivorship literature is still in its early stages, and the current meta-analysis is expected to impact the literature through the warning signals given, its results, and its focus on survivors' mental health as a priority. It is also worth noting that the meta-analysis also makes recommendations for a better assessment of the link between CRF and mental health, that of using CRF instruments that are more extensive in number of items and that measure more dimensions, as they better capture the concept. Effect sizes were found to be significantly larger for the FRC-mental health relationship where a scale exceeded five items. 

Study 3

Study 3, conducted on 72524 Reddit posts by cancer survivors, is the first study of its kind in the survivor community. The main result was that our data mining and natural language processing methods were sensitive enough to identify depression and anxiety in the content of the posts which culminated in the identification of differences between short-term survivors and long-term survivors. The former had significantly more posts labeled as relevant to anxiety and depression than long-term survivors. The results indicate that Reddit texts are an indicator of when stressors are activated (e.g., approaching the year of diagnosis) and mental health issues are triggered. Our recommendation is that the literature should pay particular attention to the mental health of short-term survivors, a result also consistent with our previous studies in this project. 

Text analysis provided the opportunity to access participants' experiences through their discourse in real time, beyond the simple recollection of events, as data is usually collected through self-reporting tools. This study contributes to the literature by using techniques that allow access to data inaccessible to mental health specialists by standard means. In future projects, the automated social media text analysis methods developed in this project can be used to detect vulnerable individuals in advanced or early stages of psychopathological problems.

These automated word processing analyses contribute to the existing literature on mental health problems experienced by cancer survivors because they use methods and a large volume of data that are inaccessible to mental health specialists by standard means. The results that short-term cancer survivors post more messages relevant to depression and anxiety than long-term survivors highlight the need for mental health screening of this population, which we are meeting through the development of the aiCARE system. 

Study 4

Study 4 introduces the aiCARE system  whose framework supports screening for mental disorders in cancer survivors and the general population. The main goal was to have a chatbot framework in which other screening tools accessible as a natural conversation with the user could be implemented. The chatbot has been implemented in the Discord platform and is openly accessible after the project is completed. Along these lines, 93% of the participants involved in the chatbot evaluation rated the Discord platform as a good hosting environment for a chatbot, with Discord being a popular platform among young adults. 

Because the chatbot can be used by anyone, is customizable with other screening tools, and is easy to access and use, it has the potential to impact primary and secondary prevention of mental health problems. Its active listening component, the summary score and the interpretation given to it, along with the natural conversation based on questions from well-validated instruments and the therapeutic coping recommendations at the end encourage its use on a larger scale. 

Study 5

The results of this study are informative regarding the possibility of using the aiCARE conversational agent, a system that incorporates artificial intelligence for natural language processing, for the purpose of assessing and screening for depressive, anxiety or post-traumatic stress symptoms. Since the results of this project show that the screening performed by the conversational agent has a high accuracy in detecting clinical symptoms as present or absent for all three screening tools that are part of the system (PHQ-9, GAD-7, PCL-5), future projects may deepen the use of the system for screening other pathologies. The chatbot in its current version does not replace clinical diagnosis, but is complementary to it. In future projects, it can be further validated by contrasting it with diagnoses made by specialists, bringing it closer to diagnostic use. 

Summary

The overall implications are that aiCARE could be used as part of a stepped care model to identify common psychopathology and to guide at-risk users to further diagnosis and care. By detecting symptoms at early stages, it can reduce the latency between early symptoms - diagnosis - treatment. aiCARE contributes to the evolving field of artificial intelligence in the early detection and diagnosis of mental health problems, shaping what future screening and treatment of mental health problems will look like.