Illusions of "all-seeing" AI in psychotherapy. Boundaries of the knowable: between body biology and digital phenotyping.
Article narration
Loading…In private practice, we regularly encounter client requests that at first glance seem purely psychological: "I have no energy", "I am constantly anxious", "I have lost interest in work and family". The standard algorithm for a non-medical psychologist or psychotherapist in such cases has long been regulated: thorough anamnesis collection, search for intrapsychic conflicts, analysis of the family system and object relations, identification of maladaptive thought patterns, or deep processing of traumatic experience. In our work, we clearly see and understand how the psyche influences the body. The concept of psychosomatics — the understanding of how repressed emotions or chronic distress lead to problems with the gastrointestinal tract, migraines, muscle spasms — has firmly entered our working arsenal and become a classic.
However, a blind spot hides in this established approach. We rarely look in the opposite direction — from the perspective of somatopsychology. We overlook how a primary change in hormonal balance, hidden systemic inflammation, or a subtle biochemical imbalance dictates the client's emotional reactions, cognitive distortions, and behavioral patterns. When a psychologist tries to work with severe apathy caused by tissue hypoxia due to iron deficiency, exclusively through cognitive restructuring or psychoanalysis methods, in essence, they are trying to heal an open fracture with conversation.
Yes, the working protocols of a responsible specialist include the point "recommend consultation with a therapist or endocrinologist", but real integration typically ends there. If a client brings test results or a medical diagnosis, most psychologists simply don't know how to integrate these objective biological data into the very fabric of the psychotherapeutic process. Psychology and medicine continue to exist in parallel universes, repeating the error of Cartesian dualism, which divided man into "thinking substance" and "extended flesh".
In this article, we will thoroughly examine how biochemical indicators and telemetry data influence differential diagnosis and the course of therapy, based on modern research. We will also consider the delicate ethical line of introducing artificial intelligence (AI) into our profession: why the integration of physiological parameters through new digital tools allows for uniting psyche and physiology in a single therapeutic space, and why Western attempts to use AI for total scanning of client emotions via video call threaten the very essence of psychotherapy.
Biochemical Landscape: What Lies Behind the Facade of Psychological Defenses
To begin, let's turn to statistics and current research. According to the World Health Organization, up to 30% of patients seeking primary care with symptoms of depression and anxiety have hidden somatic pathologies that act as the primary factor in their psychological state. Trying to treat such patients exclusively through psychotherapeutic contact — means ignoring the material substrate of their suffering.
Let's consider several key biochemical markers that are critically important to take into account in therapy:
1. Systemic Inflammation and Neuroinflammation (C-reactive protein — CRP).
In recent years, the cytokine (inflammatory) hypothesis of depression has been actively developing in global psychiatry. Authoritative studies (in particular, fundamental works by Miller A.H. and Raison C.L., as well as publications by the Scientific Center for Mental Health of the Russian Federation) convincingly show that an elevated level of high-sensitivity C-reactive protein (hs-CRP) directly correlates with so-called treatment-resistant depression.
The following occurs: cytokines penetrate the blood-brain barrier and significantly disrupt serotonin and dopamine metabolism in the brain.
What this means for a psychologist: a client with high CRP will inevitably exhibit pronounced anhedonia, psychomotor retardation, and resistance to therapy. Psychological interventions requiring high cognitive load or emotional engagement will repeatedly encounter a biological barrier — the client's brain is physiologically in a 'sickness behavior' mode.
2. Hypothalamic-Pituitary-Adrenal (HPA) Axis (Cortisol and Prolactin).
Chronic stress leads to profound dysregulation of this axis. Persistently elevated cortisol is toxic to the brain: it disrupts the function of the hippocampus (the area responsible for memory and contextualization of emotions) and hyperactivates the amygdala (the evolutionary fear center).
Furthermore, in clinical practice, there is a concept of stress-induced hyperprolactinemia. Elevated prolactin in non-pregnant women and men is often a clear marker of chronic exhaustion and presents with symptoms of pronounced emotional lability, tearfulness, and unprovoked anxiety.
What this means for a psychologist: if a client's cortisol and prolactin levels are skyrocketing, their psyche is permanently in a biological 'fight or flight' survival mode. In such a state, it is absolutely ineffective and even dangerous to work with deep-seated traumas — this is highly likely to cause retraumatization. The therapist's primary task here is grounding techniques, radical reduction of allostatic load, and regime normalization.
3. Tissue respiration and neurotransmitters (Ferritin, Vitamin D, Magnesium, TSH). Latent iron deficiency (diagnosed when ferritin is low despite normal hemoglobin) directly leads to brain tissue hypoxia.
Subjectively, clients experience this as a complete lack of motivation, a leaden heaviness in the body, and an impenetrable cognitive fog. Reduced thyroid function (even subclinical hypothyroidism, where TSH is at the upper limit of normal) physiologically slows overall metabolism, precisely mimicking classic clinical depression.
Without considering these objective data, a therapist might struggle for months with client “resistance,” while the client simply lacks the basic biological fuel needed to power the prefrontal cortex, which is critically necessary for awareness, insights, and change.
Telemetry: Beyond Subjective Self-Report
Beyond biochemistry, a crucial layer of data is telemetry — metrics continuously collected by smart wearable devices (smartwatches, fitness trackers). In the classical paradigm, we are accustomed to relying solely on the client's subjective self-report: “I slept normally,” “I don't think I was very nervous this week.” However, as professionals, we know that subjective perception is often distorted by alexithymia, repression, or psychological defenses.
Heart Rate Variability (HRV)
HRV is the variation in time between consecutive heartbeats. From the perspective of Stephen Porges' famous Polyvagal Theory, HRV is a direct and the most accurate non-invasive marker of vagal nerve tone and the balance between the sympathetic and parasympathetic nervous systems. High HRV indicates good physiological adaptability, psychological flexibility, and a high capacity for self-regulation. Low HRV is an undeniable sign of sympathetic dominance (mobilization mode) or parasympathetic rigidity (freezing/dissociation mode).
Sleep Architecture. Deprivation of deep sleep phases or fragmentation of the REM phase (rapid eye movement sleep, during which critical unconscious emotional processing of daytime experiences occurs) leads to a sharp decrease in frustration tolerance. A client may claim to have slept the required 8 hours, but objective trackers will show a complete lack of quality recovery. Access to this data between sessions allows the therapist to see the client's actual stress curve, rather than just the consciousness-censored picture the client retrospectively brings to a 50-minute weekly meeting.
The Illusion of the “All-Seeing Eye”: Why Algorithmic Emotion Recognition Destroys Therapy
For several years, international HealthTech companies have been working on the concept of "digital phenotyping" — collecting objective data from smartphones and wearable devices to predict depressive episodes or manias. And while monitoring heart rate or sleep is an undeniable benefit, other technological trends cause serious professional concern.
Today, in the US and Europe, there is a growing attempt to integrate artificial intelligence systems based on computer vision algorithms (Emotion AI, such as developments based on Affectiva or similar startups) into psychotherapy via video conferencing. These systems scan the client's face in real-time during a session, analyze micro-expressions using Paul Ekman's system, track gaze direction, pupil dilation, and display an "emotion dashboard" on the therapist's screen: "Client is experiencing anger with 82% probability," "Sincerity level is reduced."
From the perspective of depth psychotherapy and psychoanalysis, this trend is a blatant violation of the boundaries of the therapeutic process and a distortion of its fundamental essence.
First, it relies on an outdated scientific paradigm. As neuroscientist Lisa Feldman Barrett brilliantly demonstrates in her theory of constructed emotions, there are no universal "fingerprints" of emotions on the face. Furrowed brows can mean anger, intense concentration, physical pain, or a simple attempt to focus vision. Training AI to recognize "true emotions" based on facial geometry is modern algorithmic phrenology.
Second, and most importantly, reading the client's state is not a mechanical process, but a crucial part of the therapist's work, based on countertransference, empathy, and the intersubjective field. Psychoanalyst Wilfred Bion introduced the concepts of "containing" and "reverie" (the analyst's special receptivity). The therapist must feel the client's anxiety or anger with their own psychic apparatus, process it, and return it in a transformed form.
If an algorithm performs the "recognition" function, the therapist becomes alienated from the client. Instead of listening to the nuances of the contact, the specialist begins to consult a dashboard. This destroys the therapeutic alliance, turning a deep encounter of two psyches into technocratic monitoring. AI should not replace the therapist's soul; AI should not substitute empathy.
Ethical Data Integration: From Disparate Figures to a Unified Digital Phenotype
If Western facial recognition algorithms invade the space of contact, how then should technology be properly used? The answer lies in separating synchronous (live communication) and asynchronous (between sessions) collection of biological, rather than intersubjective, data. This data is more often used for medical purposes, rather than being integrated into the process of classic talk psychotherapy.
In response to this challenge, platform solutions are emerging that provide psychologists with a convenient client status “dashboard” without intruding on the sanctity of the session itself. Let's explore how this is ethically and scientifically implemented using the proEgo smart platform, which conceptually integrates several modules into a single therapist interface.
The proEgo platform aggregates four key data streams:
- Video Sessions: with transcription capability and analysis of dialogue dynamics (text analysis, not face scanning).
- Telemetry: synchronization with Apple Health/Health Connect for objective monitoring of heart rate, HRV, and sleep.
- AI Diary: a self-reflection tool between sessions, where the client can use voice input. Artificial intelligence analyzes not only text (semantics) but also speech prosody (timbre, speed, pauses, intonation), which allows for identifying acoustic markers of depressive affect flattening or anxious agitation. This occurs asynchronously, with the client's full consent.
- Biochemical Checkup: results of laboratory tests (hormones, vitamins, inflammation markers) voluntarily uploaded by the client to the system and interpreted for the therapist in terms of their influence on the psychological state.
What does this offer a psychotherapist? It creates a comprehensive, multidimensional picture. We stop working in the dark. We see when a psychological state is reinforced by biological exhaustion and can precisely adapt the strategy.
Practical Case: How It Works in the Supervision Process
To demonstrate, let's look at a real example of how the analysis of this integrated data impacts the course of therapy. We will refer to depersonalized data from the proEgo platform.
Client Context (ID 521, placeholder name Nikolay): the client is in long-term therapy (14 sessions held). The request is related to deep emotional burnout, severe future uncertainty, anxiety, and increasing difficulties in family interactions (with his son). If we worked only in a classical paradigm, we would focus exclusively on exploring parent-child relationships, searching for existential meanings, and addressing anxiety about the future.
The client maintains an AI-journal with voice messages, where the AI notes: the tone is even, sometimes sharply lowered, extraneous noises are present, the client complains about noise from the neighboring apartment during sessions, which causes him strong, uncontrollable irritation.
Somatic Status Connection: the client uploads the results of an extended lab panel. The following critical deviations are displayed in the therapist's interface:
- Cortisol: 750 nmol/L (significantly above normal, normal up to 635).
- Prolactin: 410 mIU/L (above normal).
- C-reactive protein (hs-CRP): 4.2 mg/L (systemic inflammation, normal up to 3.0).
- Ferritin: 28 µg/L (pronounced iron depletion, latent iron deficiency).
- Vitamin D: 18 ng/mL (acute deficiency).
- HOMA-IR Index: 3.9 (clear signs of insulin resistance).
AI Supervision and Correction of Therapeutic Strategy: based on session transcripts, acoustic analysis of the voice diary, and somatic status, the system generates a supervisory summary for the therapist.
What does the specialist see? The system highlights an important, previously non-obvious relationship: elevated cortisol, prolactin, and C-reactive protein are directly associated with stress states, clinical anxiety, and a high risk of emotional burnout.
In addition to psychological causes, low ferritin and vitamin D can explain complaints of chronic lack of energy and apathy. And a high HOMA-IR index indicates metabolic swings that can physiologically provoke sudden bouts of irritability and anxiety (due to fluctuations in sugar and adrenaline).
Recommendations for the Therapist (AI Supervision Conclusions): along with the classical interpretation of client irritation (for example, due to a noisy neighbor) as unconscious resistance to therapy or negative transference, supervisory analysis also recommends fundamentally different steps:
1. Validate Somatic Status: connect current experiences of total uncertainty and anxiety with check-up results. Explain to the client that their skyrocketing cortisol levels and systemic inflammation make their nervous system hypersensitive to any, even minimal, frustrations (including noise from a neighbor). This relieves the client of the toxic guilt that they are “not coping” with their life.
2. Reduce the Pace of Therapy: high stress hormone levels and critically low ferritin mean that the client currently simply does not have the physiological resources for deep transformational work or harsh confrontations. The strategy should shift towards supportive therapy, seeking external resources, and normalizing daily routines (sleep, rest).
3. Focus on self-compassion: return to matters of self-care. The client needs a clear understanding that their body is currently operating at its biological limits. This requires a deep exploration of the relationship with the body, not as a mechanical function obligated to «endure», but as a living, vulnerable organism.
Conclusion
Integrating biochemistry, telemetry, and speech analysis into psychotherapy does not make us doctors. We do not make medical diagnoses or prescribe medications. We reject dystopian ideas of total client facial scanning, preserving the inviolability of the intersubjective therapeutic field and the value of live human empathy.
However, the availability of objective biological data (provided by the client themselves) allows us to build much more precise, safe, and truly empathetic work. When a therapist sees the full picture, they stop expecting the impossible from an exhausted organism. They clearly understand where psychological defense ends and tissue hypoxia or neuroinflammation begins.
The emergence of tools such as integrated platform solutions allows us to overcome Descartes' dualism, which artificially separated body and soul. The future of psychotherapy lies precisely in this plane — in a careful, scientifically grounded, and comprehensive understanding of the individual, where blood test results and heart rate are as crucial for understanding a client's subjective well-being as their dreams, fantasies, or childhood memories.
From the editors:
As an illustration for Dana Janson's article, we chose René Magritte's iconic painting «The Therapist» (Le Thérapeute, 1937). This surrealist work is an absolute visual fit for the article's concept and brilliantly illustrates the challenge facing modern psychotherapy. The painting depicts a person whose torso is an open birdcage, inside which sit two live birds, and whose face is hidden under a wide-brimmed hat. In the context of an integrative approach, the cage represents our physical body with its biochemistry, hormonal background, and somatic status, while the birds represent our fragile psyche and emotions. A psychotherapist cannot treat only the soul while completely ignoring the state of the physiological vessel in which it resides. Magritte's painting metaphorically calls on specialists to look «inside» the framework, not limiting themselves to conversations with the faceless facade of psychological defenses, and reminds us: true healing begins where the biology of the body meets the depths of the unconscious.
References:
1. Barrett, L. F. (2017). How Emotions Are Made: The Secret Life of the Brain. Houghton Mifflin Harcourt. (Critique of the universality of facial expressions).
2. Bion, W. R. (1962). Learning from Experience. London: Heinemann. (Concept of containment and reverie).
3. Miller, A. H., & Raison, C. L. (2016). The role of inflammation in depression: from evolutionary imperative to modern treatment target. Nature Reviews Immunology, 16(1), 22-34.
4. Porges, S. W. (2011). The Polyvagal Theory: Neurophysiological Foundations of Emotions, Attachment, Communication, and Self-regulation. W. W. Norton & Company.
5. Insel, T. R. (2017). Digital Phenotyping: Technology for a New Science of Behavior. JAMA.
6. Uspenskaya, Yu. A., et al. (2019). The Role of Systemic Inflammation in the Pathogenesis of Depressive Disorders (Mental Health Research Center of the Russian Federation).
7. McWilliams, N. (2004). Psychoanalytic Psychotherapy: A Practitioner's Guide. Guilford Press.
Комментарии аккредитованных экспертов (0)
Войдите, чтобы оставлять комментарии
