Digital and cyber psychiatry: patient transformation in the digital age

Full Text

Introduction

We live in an era in which the average person globally spends more than six hours per day in a digital environment [1]. Unlike the natural or industrial environment, this one was created not by nature or machinery, but by our own minds. And that same mind is beginning to malfunction. The paradox of the present moment is that the more deeply we study the harmful effects of the internet, social media, and algorithms, the more sophisticated our digital tools for diagnosis and therapy become. We find ourselves in the position of a physician treating poisoning with the very substance that caused the intoxication, only in a different dose. It is precisely this dialectical tension between cyber and digital psychiatry that we wish to explore, moving beyond the conventional boundaries of a traditional review article and engaging instead in a more reflective and critical discussion.

Two poles of a single ecosystem

To avoid drowning in terminology, let us first map the conceptual landscape.

Cyber psychiatry conceptualizes the digital world as a pathogenic factor. It examines how informational chaos and the predominance of simulacra generate anomie (the breakdown of social norms and values), chronic distress, and depersonalization [2]. By the dominance of simulacra, we mean the replacement of objective reality with constructed digital images. This may contribute to social disorientation (anomie) and an increase in dissociative states in which individuals lose touch with the physical context of their existence. From a neurobiological perspective, the smartphone functions as an exceptionally effective operant conditioning device: a variable reinforcement schedule (“Did I get a like or not?” or “Is this message important or spam?”) engages dopaminergic pathways with an efficiency that would have impressed any experimental psychologist of the mid-20th century [3].

Digital psychiatry, by contrast, views technology as a therapeutic resource. Telepsychiatry became a standard model of care following the COVID-19 pandemic [4], virtual reality environments that allow exposure therapy for phobias without direct real-world risk [5], and agentic artificial intelligence (AI) has the potential to become an assistant that does not experience fatigue and may reduce errors during routine screening [6].

The challenge is that these two domains are inextricably linked. A tool designed to provide assistance may unintentionally amplify the very risks it was intended to mitigate. This is the “digital paradox” of psychiatry.

The evolution of diagnosis: from addiction to anomie and cognitive epidemic

Tracing the development of our understanding of digital addictions helps to illuminate how this pathogenic process has evolved over time. A quarter century ago, Kimberly Young adapted the criteria for pathological gambling to internet addiction [7]. Later, as digital ecosystems expanded, new phenomena emerged, including nomophobia (the fear of being out of mobile phone connectivity) [8], FoMO (fear of missing out online) [9], and social media addiction [1]. The culmination of this process was the inclusion of gaming disorder in the International Classification of Diseases, 11th Revision (ICD-11) [10]. The core diagnostic features remained unchanged: loss of control, increasing prioritization of digital activity over real-world activities, and continued use despite negative consequences.

However, the discussion today extends far beyond behavioral addictions. In recent years, we have witnessed the growing recognition of what might be termed the social psychopathology of the digital age. Russian researchers have documented an increase in irrational psychological defenses: ways of escaping an anxiety-provoking reality through primitive, comforting, yet false explanations, ranging from conspiracy theories to pseudo-expert phenomena. This latter phenomenon is itself dialectical: online pseudo-experts do not merely promote useless, simplistic solutions — consumers themselves willingly pay for the “thrills of magic pills”, thereby further contributing to large-scale neuroticization [2]. The unprecedented speed and reach of digital misinformation and disinformation create a burden that the human cognitive system, adapted to slow information exchange, is incapable of handling. This leads to an “explosion” of irrational defenses [11]. We are no longer dealing with a new “LLM wrapper”, but with a transformation of the mental environment itself, in which the scale and speed of algorithmically amplified psychological contagion give rise to an informationally conditioned cognitive epidemic (“informational neurosis”, according to Khananashvili, 1978) [12].

Even more concerning are findings suggesting that intensive use of generative AI models is associated with reduced critical thinking performance among students [13]. We risk raising a generation that places greater trust in algorithms more than its own reasoning. And here we approach a main methodological trap — reification.

Risks: reification, or when the algorithm begins to believe in its own objectivity

The philosophical term “reification” refers to turning something abstract as if it were a concrete entity. In the context of digital psychiatry, this occurs when a complex psychiatric construct imbued with cultural and personal significance (for example, “depression” or “suicide risk”) is reduced to a simplified numerical index derived from smartphone accelerometer data or voice analysis [14]. AI does not understand context, yet it presents its outputs with spurious precision. This creates a highly dangerous illusion of objectivity.

Consider the following situation: a psychiatrist, relying on an “objective” AI-generated suicide risk score (for example, 0.12), ignores the patient’s nonverbal indicators of distress (tremor, avoidance of eye contact). At that moment, the fiduciary bond begins to erode: the psychiatrist ceases to be an understanding subject and instead becomes the operator of a device, while the psychiatric encounter is reduced to a technical measurement, thereby undermining the meaning of the therapeutic process. The unique experience of a struggling individual is reduced to a set of numerical indicators. We risk creating a situation in which a psychiatric label assigned by an impersonal algorithm acquires an appearance of scientific legitimacy, thereby reinforcing stigmatization [15]. For this reason, claims that AI will soon replace psychiatrists are not merely naive, but potentially dangerous. The solution lies not in banning technology, but in fostering critical evaluation and mandatory specialist verification of AI-generated outputs.

Operational governance: the uncertainty of the digital future

At international congresses, such as the Digital Mental Health Global Congress 2025 in Toronto, a growing consensus has emerged that the primary challenge in digital psychiatry is not technological limitations, but the absence of effective governance mechanisms [16]. The Australian Psychological Society has already raised concerns regarding the risks of entrenching inequality and creating new ethical challenges through the unregulated implementation of large language models [17]. For example, unregulated AI may perpetuate stigma by generating harsher predictions for certain social groups due to biases in training datasets.

This inevitably raises several simple questions for which no clear regulatory solutions currently exist:

• Who bears ultimate responsibility for AI-assisted decisions in everyday clinical practice?
• Who has the authority, and on what grounds, to override an AI-generated decision?
• If the algorithm makes an error and harms a patient, who stands trial: the software developer, the medical device validator, or the treating physician?

Until we establish clear regulatory and ethical frameworks, the implementation of agentic AI in psychiatry will resemble flying an airplane without either a pilot or a parachute.

Blended care models: keep “human in the loop”

A resolution to this contradiction lies neither in techno-optimism nor in neo-Luddite skepticism, but in a balanced middle ground. This approach is reflected in the so-called blended-care models, which integrate digital tools into face-to-face clinical practice [18]. Self-help CBT (cognitive behavioral therapy) interventions may be used to monitor mood between sessions, virtual reality may prepare patients for challenging social situations, and telepsychiatry may expand access to care in remote areas. However, the final interpretation of the data, responsible clinical judgement, and, most importantly, the empathic human connection must remain in the hands of the clinician.

For clarity, Table 1 summarizes the risks and prospects associated with these two domains in a way that better reflects clinical reality than formal evidence hierarchies.

 

Table 1. Risks and prospects of cyber and digital psychiatry

Aspect	Cyber psychiatry	Digital psychiatry
Objects of focus	Pathogenic mechanisms of digital environment	Diagnostic and therapeutic tools
Key phenomena	Anomie, depersonalization, loss of critical thinking	Telemedicine as a human-centered approach; Digital phenotyping and agentic AI as potentially dehumanizing approach
Major risks	Algorithm-driven formation of “suicidal funnel”*	Reification** — replacement of clinical judgment with algorithmic calculation
Solutions	The implementation of critical thinking development programs and cognitive remediation as a primary preventive strategy against digital psychopathology	Risk management and the development of blended care models with a real rather than merely nominal “human in the loop”

Note: *Content recommendation algorithms that, after detecting a user’s interest in self-harm–related material, begin to progressively suggest similar depressive content within a positive feedback loop. **The transformation of mental phenomena into measurable entities, reducing subjective suffering to a digital object.

 

Based on the available literature, we propose the following roadmap for the implementation of digital technologies in psychiatry, which synthesizes existing approaches to risk management and the phased adoption of innovation:

1. Preclinical validation stage:
   
   • development of standards for clinical trials of AI-based tools;
   • validation using independent clinical samples;
   • multidisciplinary expert evaluation (psychiatrists, ethicists, engineers).
2. Pilot implementation stage:
   
   • establishment of clear use protocols;
   • assignment of personnel responsible for oversight (“a human in the loop” with clearly defined authority);
   • collection of safety and efficacy data.
3. Scaling stage:
   
   • development of specialist training programs;
   • establishment of post-marketing monitoring systems involving independent professional associations (to prevent monopolization and conflicts of interest);
   • regular ethical audits.
4. Institutionalization stage:
   
   • incorporation of AI-assisted clinical decision-making methodologies into clinical guidelines and standards of care;
   • integration of AI-related competencies into continuing medical education systems;
   • development of insurance and financial models.

Conclusion

The digital transformation of psychiatry is inevitable — it is neither inherently good nor inherently harmful, but rather an objective reality. However, the pace of this transformation must not outstrip our capacity for critical reflection. We must remember that psychiatry is not only the science of the brain, but also an art of understanding the individual. No matter how advanced, agentic AI cannot replace the therapeutic alliance built on trust and empathy. The future of the discipline lies not in replacing psychiatrists with machines, but in equipping them with digital tools that augment their capabilities without diminishing responsibility or undermining clinical reasoning. This is not the time for uncritical acceleration, but to pause and reflect on who will govern this complex digital mechanism, and how.

 

Authors’ contribution: All the authors made a significant contribution to the article, checked and approved its final version prior to publication.

Funding: The research was carried out without additional funding.

Conflict of interest: The authors declare no conflicts of interest.

Generative AI use statement: In the preparation of this work, generative artificial intelligence was used in a limited manner. Specifically, ChatGPT and DeepSeek were employed to optimize the wording of individual sentences and to check the stylistics of the Russian-language text after the main stage of the scientific work had been completed. All AI-generated fragments were carefully reviewed and edited by the authors to preserve scientific accuracy and align with the original meaning. The final responsibility for the content of this article, including the interpretation of data and conclusions, lies entirely with the author team.

 

About the authors

Nikolay G. Neznanov

V.M. Bekhterev National Medical Research Centre for Psychiatry and Neurology

Email: goncharenko7@yandex.ru
ORCID iD: 0000-0001-5618-4206
SPIN-code: 9772-0024
Scopus Author ID: 35593613200
ResearcherId: U-1562-2017

MD, Dr. Sci. (Med.), Professor, Director, Chief Non-staff Specialist Expert in Psychiatry of Roszdravnadzor, Chairman of the Board of the Russian Society of Psychiatrists

Russian Federation, Saint Petersburg

Natalia V. Semenova

V.M. Bekhterev National Medical Research Centre for Psychiatry and Neurology

Email: goncharenko7@yandex.ru
ORCID iD: 0000-0002-2798-8800
SPIN-code: 3552-1894
Scopus Author ID: 16639990900
ResearcherId: I-1030-2018

MD, Dr. Sci. (Med.) Deputy Director for Scientific, Organizational and Methodological Work

Russian Federation, Saint Petersburg

Andrey Yu. Goncharenko

V.M. Bekhterev National Medical Research Centre for Psychiatry and Neurology

Author for correspondence.
Email: goncharenko7@yandex.ru
ORCID iD: 0000-0001-5208-4099
SPIN-code: 1193-8221
Scopus Author ID: 8540267600

MD, Dr. Sci. (Med.), Leading researcher, Scientific Organizational Department

Russian Federation, Saint Petersburg

References

Supplementary files