Artificial Intelligence in Neuropsychiatry: A Potential Beacon in an Ocean of Uncertainty?

By the moment I am writing these lines, 2023 is coming near its end. It seems a certainty that 2023 will undoubtedly be remembered as the year when Artificial Intelligence (AI) became an integral part of our lives. The introduction of user-friendly language models like ChatGPT, Bing, and Bard, along with the development of AI-based smartphones, marks a significant shift. AI is not just a technological advancement; it's becoming our "newest artificial extension," akin to how early humans started crafting tools at the dawn of humanity. This evolution is now extending into the realms of medicine and neuropsychiatry, showcasing AI's growing influence in these critical fields. In particular, the integration of AI into neuropsychiatry signifies a pivotal shift in the medical field, merging cutting-edge technology with clinical expertise. AI's proficiency has been demonstrated in various domains of healthcare, notably in psychiatry where it aids in the early detection of diseases, comprehends disease progression, and uncovers new treatment methods. In particular, AI's role in diagnosing psychiatric disorders such as depression, schizophrenia, autism spectrum disorder, and Alzheimer's disease is increasingly recognized. Its application extends to real-time monitoring and intervention strategies, enhancing the scope and effectiveness of psychiatric care. AI's contribution to psychiatry also includes its ability to synthesize and analyze vast datasets, providing insights into complex psychiatric conditions. This capability is instrumental in accelerating the diagnostic process, identifying key indicators of various mental health issues. The use of AI in psychiatric diagnosis and therapy, particularly in conditions like Alzheimer's disease and mild cognitive impairment, has been validated through various methods including neuroimaging and neuropsychologi-cal data analysis [https://pubmed.ncbi.nlm.nih.gov/37506575/].

Additionally, the advancement in AI-driven technologies has catalyzed a new approach known as "NeuroAI," which aims to develop AI systems with advanced sensorimotor abilities akin to humans and animals. This approach, rooted in the concept of the "embodied Turing test," expands the scope of AI applications beyond traditional domains, focusing on the interaction of artificial systems with the physical world [https://www.nature.com/articles/s41467-023-37180-x]. Despite these promising developments, AI in psychiatry faces challenges related to training data quality and ethical considerations. It is crucial to view AI as a complementary tool to human expertise, rather than a replacement, to enhance the diagnostic and therapeutic processes in mental healthcare. Indeed, AI's integration into neuropsychiatry opens a new frontier in medical research and practice. Its evolving role in understanding, diagnosing, and treating psychiatric disorders makes it an indispensable tool in the field, promising to reshape the future of mental healthcare.

On the other hand, the progressive fusion of AI with medicine and neuropsychiatry, while ushering in innovative advancements, also hides complex challenges. Ethical considerations in the patient-caregiver dynamic are at the forefront, as AI's growing role could potentially alter trust and decision-making processes. AI's effectiveness in diagnosing and treating mental health disorders is pivotal, yet it raises questions about maintaining human-centric approaches in therapy and care. Indeed, such a rapid evolution raises critical questions: “Could our growing reliance on AI lead to new forms of dependency, such as an over-reliance or emotional attachment to AI systems?” These concerns, touching on potential issues like AI dependency or idolization, are especially pertinent as we consider the growing expansive influence of AI in everyday life. Furthermore, this apparently unavoidable shift might even potentially re-shape (if not exacerbate) pre-existing psychiatric conditions, leading to new dimensions of disorders. For example, in individuals with depression, AI might intensify feelings of isolation and loneliness. For those with schizophrenia, the technology could blur the lines between reality and delusion. Anxiety disorders might see heightened symptoms due to over-reliance on AI for decision-making. Obsessive behaviors in OCD could be triggered by the predictability and precision of AI. In ADHD, the constant stimulation from AI interactions could exacerbate attention difficulties. Finally, for individuals on the autism spectrum, AI might become a preferred mode of interaction, impacting social skills development [Mucci, Vannini, Mennucci 2024, in preparation].

In this scenario of potentially epochal change, this bimonthly issue of Clinical Neuropsychiatry opens with an editorial dedicated to this hot topic that increasingly seems doomed to become a real social plague on a global scale, namely problematic Internet use. In his insightful editorial, Adriano Schimmenti [pp. 471-478] challenges the traditional views by suggesting that such problematic behaviors are deeply linked to an individual's basic needs and personal history, anchored in the appetitive nature of their online interactions. This perspective reimagines the Internet as an environment where users engage to fulfill distinct needs, rather than just a tool for various actions. Schimmenti advocates for a reassessment of how excessive Internet use is categorized, cautioning against its immediate classification as a behavioral addiction. The article emphasizes a thorough evaluation of individuals displaying dysregulated Internet behaviors, concentrating on the longevity and significance of these behaviors, as well as the underlying motivations. This approach enriches the understanding of symptom development and highlights the importance of recognizing personal motivations in online behavior, which is crucial for developing personalized treatment strategies and improving case assessments. As a matter of facts, an escalation of problematic internet use during the COVID-19 pandemic has been highlighted in dozens of recent studies, underscoring the profound impact of prolonged home confinement and altered daily routines on mental health. Such a rise in digital engagement, while serving as a coping mechanism for many, also brought to the fore significant mental health concerns. In such light, the study conducted by Orrù and colleagues [pp. 505-510] offers a timely investigation into another facet of mental health during the pandemic – namely the subjective memory complaints. This shift from the digital to the cognitive impact of COVID-19 paints a comprehensive picture of the pandemic's multifaceted influence on psychological well-being. Acknowledging existing research that links SARS-CoV-2 infection with memory impairments, the study specifically investigates how COVID-19 may affect individuals' perceptions of their memory capabilities, in the absence of objectively measurable memory loss. Utilizing self-reported assessment tools, namely the "Subjective Memory Complaints Questionnaire" (SMCQ) and the "Prospective and Retrospective Memory Questionnaire" (PRMQ), the study compared participants' memory perceptions during two distinct periods: pre-pandemic and during the pandemic. The research involved 207 Italian participants, predominantly females aged between 55 and 65 years, out of which 189 were included in the final analysis. Results indicated a significant increase in total PRMQ scores during the pandemic period compared to the pre-pandemic period. However, no notable differences were observed between COVID-19-positive individuals and those who had not contracted the virus in terms of PRMQ and SMCQ scores. A detailed analysis using McNemar's test revealed significant increases in specific items of the SMCQ, suggesting a perceived deterioration in memory function among participants during the pandemic.

Orrù’s investigation of cognitive changes during the pandemic sets the stage for a deeper dive into the broader mental health ramifications, as explored by Glavina and his team [pp. 479-485. Specifically, the primary aim of their study was to investigate the prevalence of mental disorders in these patients and identify risk factors contributing to deteriorating mental health during their hospital treatment. This research involved 135 patients discharged in January 2022, with a comprehensive collection of their sociodemographic data, somatic comorbidities, and treatments received during hospitalization. A significant focus was placed on monitoring psychiatric diagnoses, therapy, and the initiation of psychotropic drug use, both at the time of hospital discharge and one year later. The findings revealed a notable increase in the use of psychotropic drugs, excluding antidepressants, during hospitalization for COVID-19. There was a substantial rise in the use of anxiolytics, antipsychotics, and hypnotics, with these numbers returning closer to baseline after discharge. The authors conclude that hospitalization for COVID-19 is associated with a marked decline in mental health, potentially linked to a fear of death, which aligns with the "landscape of fear" theory.

Transitioning from the in-depth analysis of COVID-19's psychological impacts, the next paper presented in this issue directs our focus towards the complex issue of food addiction in individuals consulting nutritionists. Piccinni and colleagues [pp. 486-494] provide a critical examination of the prevalence and characteristics of food addiction, a condition characterized by the compulsive consumption of high-energy foods. The study, involving 842 participants, seeks to determine the prevalence of FA and its association with eating habits. Using a self-administered questionnaire and the Yale Food Addiction Scale (YFAS) for assessment, the research conducted various statistical analyses, including Chi-square and independent t-tests, along with a stepwise logistic regression model that accounted for age, sex, and body mass index (BMI). The findings reveal that 15.3% of the sample displayed FA, with similar rates between women and men. However, FA was more prevalent among overweight individuals. Women, as per YFAS, tended to report more persistent desires and withdrawal symptoms than men. Those diagnosed with FA were found to have made more attempts at weight loss, engaged in self-dieting, had a varied mealtime routine, and tended to snack continuously throughout the day. Notably, carbohydrate consumption in a single meal was significantly correlated with FA. The research concludes that individuals seeking dietary advice from nutritionists often exhibit characteristics of FA. Despite diagnostic debates surrounding FA, its potential role in obesity and other psychopathological conditions warrants further investigation and tailored therapeutic interventions.

Notably, recent studies have found a significant association between eating disorders (and possibly FA) and autism spectrum disorder (ASD). These studies highlight that individuals with eating disorders often show ASD-like traits, and abnormal eating behaviors are commonly seen in those with ASD. This intersection of psychopathological traits leads us to the work of Dell'Osso and team [pp.505-510], who present an update of their previous, unique, case-report exploring the potential neurodevelopmental trajectory of Marilyn Monroe's mental health. While their study on Marilyn Monroe doesn't explicitly mention FA, Monroe's known eating behaviors suggest a possible connection to eating disorders. However, Dell'Osso's analysis provides an insightful exploration of the intersection of (ASD) [or autistic traits (Ats]) with borderline personality disorder (BPD) and bipolar disorder (BD), highlighting the continuum nature of ASD and ATs in both clinical and general populations. This analysis particularly focuses on how ASD may have played a role in both Monroe's extraordinary success and her challenging mental health journey. Utilizing a variety of sources, including her biographies, filmography, interviews, and personal scripts, the study constructs a hypothetical illness trajectory for Monroe, starting from ASD and evolving into BD with unspecified catatonia, culminating in her untimely and unresolved death. This case analysis underscores the significant impact of psychological traumas on individuals with ASD vulnerability and its correlation with the development of BPD and BD. The research positions Marilyn Monroe as a potential prototypical example of BPD from a neurodevelopmental viewpoint, suggesting a progression to BPD and BD with catatonia, exacerbated by multiple traumas. This perspective aligns with the DSM-5-TR (2022) descriptions of high-functioning autism in verbally adept adults without intellectual impairment.

This issue of Clinical Neuropsychiatry concludes with an investigation conducted by Pedone and Semerari [pp. 471-478] that aims at introducing the Metacognition Brief Rating Scale (MBRS), namely a novel informant tool designed as a complement to the Metacognition Self-Assessment Scale (MSAS). The MSAS, originally developed to assess metacognition - the ability to understand one's own and others' mental states, including emotions, beliefs, and intentions, raised concerns regarding the accuracy of self-reports by individuals who might lack these metacognitive abilities. To address this, the MBRS was created, transforming MSAS items into a third-person perspective. The research involved administering the MBRS to a sample of 384 randomly selected individuals from the general population. The study was structured in three methodological steps: rewriting MSAS items for the MBRS, testing the congruence of the four-factor structure between MBRS and MSAS through exploratory and confirmatory factor analysis, and evaluating and comparing the psychometric properties of both scales. The findings indicated that both MBRS and MSAS possess adequate psychometric properties, with a validated four-factor structure for the MBRS. A significant correlation was observed between the two versions, suggesting similar factor structures. The authors conclude that the MBRS holds promising psychometric properties, but emphasizes the need for further research in larger and varied samples to confirm and expand these findings.