Why Can't We Do Psychiatry as a System?

As mental healthcare becomes more technologically advanced, the challenge is to connect clinical intelligence with the lived reality of each patient. Dr. Thalhath P, MBBS, MD Psychiatry Chief Psychiatrist and Managing Director, Metro Mind Centre for Neuropsychiatry

A person seeking psychiatric care does not arrive as a diagnosis. They arrive with a life: experiences they may struggle to articulate, difficulties that others may not recognise, treatments that may or may not have helped, and questions about what happens next. Understanding that life is among the most important tasks in psychiatry. Yet, increasingly, we must ask whether our systems are equipped to do justice to its complexity.

For years, one question has stayed with me: why can't psychiatry bring information, technology and different disciplines together as other advanced areas of medicine do?

Psychiatry has accumulated many tools to understand, analyze and intervene the human mind: diagnostic frameworks, medication, psychotherapy, psychological assessments, brain investigations, genetic testing and neuromodulation. Artificial intelligence offers further possibilities for organising information and examining clinical patterns.

But how effectively do we bring these sources of knowledge together to understand an individual?

Consider two people diagnosed with depression. One struggles with persistent anxiety, disturbed sleep and impaired concentration. The other experiences emotional dysregulation, difficulties in relationships and repeated treatment failures. The diagnosis may be identical, but their experiences, contributing factors and clinical needs may be profoundly different.

The question is not simply whether we possess enough information. It is whether we can connect what we know with what a person is actually experiencing.

This question has shaped two initiatives at Metro Mind: Utharam-an approach to integrated mental healthcare, and ILM, Integrated Learning in Mental Health-an emerging clinical intelligence initiative designed to help clinicians interpret patient-specific information more systematically. The ambition is to approach psychiatry as a connected system of understanding.

Beyond the medical record

A psychiatric assessment is not a single event. It is an evolving process of understanding a person over time.

Clinicians collect histories, conduct mental status examinations, administer psychometric assessments, review treatments and monitor outcomes. Where appropriate, laboratory investigations, quantitative electroencephalography (qEEG) and genetic or pharmacogenomic information may add evidence.

Yet collecting information is only the beginning.

Attention difficulties, for example, may be associated with sleep disruption, anxiety, depression, medication effects or a neurodevelopmental condition. When treatment does not produce the expected improvement, the clinician may need to reconsider the diagnosis, examine previous interventions, assess adherence and investigate other contributing factors.

These questions become harder when information is scattered across records, departments and assessment systems.

Digitising a medical record does not automatically make it intelligent. A system can preserve thousands of observations without helping a clinician understand how they relate to one another.

The next challenge for psychiatry is therefore not simply digitisation, but integration: creating systems that help clinicians examine evidence across a patient's clinical journey, recognise missing information and consider alternative explanations.

What should clinical intelligence do?

ILM -- Integrated Learning in Mental Health -- is an attempt to move psychiatry from fragmented clinical information to integrated clinical intelligence. By connecting evidence across a patient's clinical journey, ILM aims to reveal patterns, challenge assumptions and bring greater precision to clinical reasoning, enabling a deeper understanding of the individual and more informed treatment decisions.

A general-purpose AI model may explain depression or describe established treatments. Patient-specific reasoning requires more: the system must account for the patient's history, distinguish past observations from current findings, examine outcomes and recognise insufficient evidence.

If a patient has not improved, a generic explanation of treatment-resistant depression is unlikely to be enough. The relevant questions are specific: What treatments have been tried? What changed after each intervention? Were assessments conducted consistently? Are there contradictory findings? What other explanations deserve consideration?

ILM is being developed around six principles: Collect, Check, Connect, Challenge, Cognition and Comprehension.

Collect means bringing relevant information together while preserving its source, context and chronology.

Check means examining the quality and consistency of that information.

Connect builds on what has been collected and checked, bringing verified clinical information together to identify meaningful relationships between symptoms, psychological assessments, biological observations, treatments and outcomes.

Challenge means questioning assumptions rather than simply reinforcing an initial hypothesis.

Cognition involves reasoning across available evidence to generate and compare possible explanations.

Comprehension is the ultimate objective: developing a coherent understanding of the individual, including what the evidence supports, what remains uncertain and what needs further investigation.

These distinctions matter: correlation is not causation, a hypothesis is not a diagnosis, and a sophisticated output is not necessarily a sound clinical conclusion.

The purpose of artificial intelligence in psychiatry should not be to manufacture certainty, but to help clinicians navigate uncertainty more rigorously.

When different kinds of evidence meet

Integrated clinical intelligence becomes especially relevant when different sources of evidence can be considered together.

At Utharam, we are exploring how psychiatric and psychological assessments can be examined alongside qEEG and genetic or pharmacogenomic investigations where appropriate.

qEEG provides quantitative information about electrical activity recorded from the scalp. It can contribute to clinical assessment, but it is not, by itself, a diagnostic test for most psychiatric disorders. Genetic information may offer insights into biological vulnerability, while pharmacogenomic testing can inform selected medication decisions. Neither can explain every aspect of a person's illness or predict treatment response with certainty.

The challenge is to interpret these findings in context, without allowing one measurement to dominate.

A person experiencing persistent attention difficulties may need an examination of developmental history, sleep, mood, anxiety, substance use, medication exposure and functioning in everyday life. Standardised assessments may help quantify the difficulties; further investigations may be appropriate when indicated.

An integrated system can organise evidence, flag relationships worth investigating and highlight what remains unknown. It should not encourage unnecessary testing or replace clinical judgement.

The same principle applies to treatment. Repetitive transcranial magnetic stimulation (rTMS), for example, has established applications in selected conditions, particularly depression. Its appropriate use depends on clinical indication, patient selection, treatment planning, monitoring and assessment of outcomes.

Emerging treatments, including ketamine-based interventions, also require careful selection, appropriate monitoring and adherence to applicable clinical and regulatory requirements.

Technology becomes valuable when it strengthens the reasoning behind these decisions and helps clinicians evaluate what happens afterwards.

Metro Mind Centre for Neuropsychiatry

Learning from every clinical journey

One of the most important opportunities in this approach lies in learning systematically from clinical outcomes.

When a treatment is prescribed, we should record not only what was done, but why it was chosen, what outcome was anticipated and what actually happened. Consistent baseline assessments and follow-up measurements can help clinicians investigate patterns, examine unexpected responses and identify questions for further research.

Such observations do not automatically establish that a treatment caused an improvement or that a particular characteristic predicts response. Those conclusions require appropriate research methods, adequate data and independent validation.

Nevertheless, meaningful clinical learning is difficult when information is incomplete, inconsistent or disconnected.

ILM is an attempt to create a foundation for more systematic learning. Rather than treating the medical record solely as a repository of past events, the system is intended to help clinicians examine those events in relation to one another.

Over time, this approach could support better research questions and more rigorous evaluation of clinical practice. Its value must be demonstrated, not assumed from the technology itself.

The clinician must remain at the centre

In discussions about advanced healthcare technology, the person can disappear behind the measurements.

Psychiatry depends on listening, trust and the ability to understand experiences that may be difficult to express. A score cannot fully capture the meaning of a relationship, the weight of a personal loss or the circumstances in which symptoms emerge. A brain investigation cannot replace a conversation about how someone experiences their own life.

This is why the growing emphasis on lived experience in mental healthcare matters. Listening to patients should be more than a preliminary step before professional interpretation begins. Their accounts should help shape the questions clinicians ask, the care they provide and the ways in which outcomes are evaluated.

For technology to contribute meaningfully, it must help clinicians attend more carefully to those experiences, not obscure them beneath increasingly elaborate datasets.

AI can overlook context or produce unsupported conclusions. Clinical responsibility must remain with qualified professionals working with patients. Privacy, informed consent, information security and transparent evaluation are essential.

An intelligent system must earn trust through its safeguards and demonstrated clinical value, not through the sophistication of its language or algorithms.

A different way forward

When we began asking why psychiatry could not be approached as a system, we were not looking merely for another software product. We were questioning how clinical knowledge is organised, connected and applied.

Utharam represents our effort to bring different aspects of mental healthcare together in practice. ILM represents our attempt to develop a clinical intelligence layer that can help clinicians examine the relationships between them.

Both initiatives remain under development. Their promise depends on scientific validation, responsible implementation and evidence that they improve clinical reasoning and patient care.

Psychiatry will continue to need skilled clinicians, compassionate listening and experience. It will also need systems that help clinicians navigate modern medicine's complexity.

The future may not be defined by a single test, a single treatment or an ever more powerful AI model. It may depend on how effectively we connect what we know with what patients tell us, recognise what we do not know and learn from what happens to each person.

The real measure of technological progress in psychiatry will not be how much information a system can process, but how much better it helps us understand the human being behind that information.

Dr. Thalhath P, Chief Psychiatrist and Managing Director, Metro Mind Centre for Neuropsychiatry.

"This is a company press release that is not part of editorial content. No journalist of The Hindu was involved in the publication of this release."