Psychiatric diagnosis is often criticised for lacking the laboratory certainty available in some other branches of medicine.

There is no blood test that independently establishes major depressive disorder, no MRI that confirms adult ADHD, and no EEG signature that by itself diagnoses bipolar disorder.

Yet this does not mean psychiatric diagnosis has to remain purely subjective.

The better direction may be to make diagnosis multimodal, structured, longitudinal and explicitly probabilistic.

I propose a practical conceptual framework for this approach:

Multimodal Diagnostic Triangulation in Psychiatry — MDTP

The central idea is simple:

Diagnostic confidence should increase when multiple independent sources of evidence converge on the same clinical formulation.

Instead of asking whether one interview, questionnaire, cognitive test, EEG or biomarker can “prove” a psychiatric diagnosis, MDTP asks a different question:

How much independent evidence supports this diagnosis, and how much evidence contradicts it?

This approach may provide a more useful pathway towards improving diagnostic accuracy in everyday psychiatry.

MDTP, as described here, is a proposed clinical reasoning framework rather than a currently validated diagnostic instrument.

Why Do We Need a Different Approach?

Psychiatric diagnosis remains challenging for several reasons.

Symptoms overlap extensively between disorders.

Poor concentration can occur with ADHD, depression, anxiety, sleep deprivation, bipolar disorder, substance use, medication effects and medical illness.

Irritability can occur in depression, bipolar disorder, ADHD, autism, personality disorders, trauma-related conditions and numerous environmental situations.

Fatigue may be psychiatric, medical, sleep-related or medication-induced.

Even structured interviews cannot completely eliminate uncertainty.

A major 2026 systematic review and meta-analysis published in JAMA Network Open examined 57 studies of standardized diagnostic interviews. The pooled test–retest reliability was approximately κ = 0.69, but reliability varied substantially between disorders and studies. The authors concluded that structural standardisation alone may not be enough and emphasised the importance of contextual and phenomenological information.

That finding points towards an important principle:

The future of psychiatric diagnosis is unlikely to be simply “more checklists.”

We need structured assessment—but we also need context.

From Diagnostic Labels to Diagnostic Evidence

Traditional psychiatric documentation often ends with something like:

Diagnosis: Major Depressive Disorder

or

Diagnosis: ADHD

But this hides an important question:

How certain are we?

A more informative formulation might say:

Adult ADHD — probable, moderate diagnostic confidence; childhood symptoms supported by retrospective history but school records unavailable; significant current functional impairment demonstrated; mood, anxiety and sleep disorders assessed; collateral history awaited.

This tells another clinician considerably more.

MDTP therefore separates two concepts:

Diagnosis

What disorder best explains the presentation?

Diagnostic confidence

How strongly does the available evidence support that diagnosis?

This distinction is especially useful in complex psychiatry.

The Eight Layers of Multimodal Diagnostic Triangulation

A comprehensive psychiatric assessment can be conceptualised as eight interacting evidence streams.

Layer 1 — Phenomenology

Everything begins with the clinical interview.

Before assigning a diagnostic label, the clinician should establish exactly what the patient is experiencing.

For example, when someone says:

“My thoughts are racing.”

Does this mean:

• rapid associative thinking during hypomania?
• repetitive anxious worry?
• obsessive intrusive thoughts?
• ADHD-related distractibility?
• trauma-associated hyperarousal?
• simply having several concerns simultaneously?

The words may sound identical.

The psychopathology may be completely different.

Good psychiatric diagnosis therefore begins with precise phenomenology rather than immediate categorisation.

Layer 2 — The Longitudinal Timeline

Psychiatric diagnosis becomes substantially more accurate when symptoms are placed in time.

Instead of asking only:

“What symptoms do you have?”

ask:

“When did they begin, what happened before them, how have they changed, and when have they disappeared?”

A useful clinical timeline is:

Premorbid functioning

↓

Earliest symptoms

↓

Age at onset

↓

Precipitating factors

↓

Evolution

↓

Episodes and remissions

↓

Treatment exposure

↓

Response

↓

Current state

This is particularly important when differentiating:

ADHD from secondary inattention;

unipolar depression from bipolar disorder;

personality traits from episodic illness;

chronic anxiety from trauma responses;

and stable cognitive limitations from progressive neurocognitive decline.

A photograph tells us what the patient looks like today.

A timeline tells us what disease process may be occurring.

Layer 3 — Collateral Evidence

One of the limitations of psychiatric diagnosis is that clinicians frequently depend on retrospective self-report.

But memory is imperfect.

Insight may be impaired.

Symptoms may appear differently to family members than to patients.

Hypomania may be remembered as:

“That was actually the most productive period of my life.”

Childhood ADHD may be difficult for a 35-year-old patient to reconstruct accurately.

Cognitive decline may be much more apparent to a spouse than to the person experiencing it.

MDTP therefore treats collateral information as an independent diagnostic stream.

Potential sources include:

• parents
• spouse or partner
• siblings
• teachers
• school reports
• previous psychiatric records
• prescriptions
• hospital records
• occupational history
• previous psychological assessments

Concordance between independent sources should increase confidence.

Major discrepancies should trigger further investigation rather than being ignored.

Layer 4 — Structured Diagnostic Assessment

Clinical intuition is important.

But clinicians are vulnerable to cognitive biases.

These include:

anchoring — becoming attached to the first diagnosis;

confirmation bias — noticing evidence that supports it;

availability bias — diagnosing what we have recently encountered;

and

premature closure — stopping once a plausible explanation appears.

Structured or semi-structured assessment can reduce some of these problems.

Tools such as the MINI, SCID and disorder-specific structured interviews force clinicians to systematically examine diagnostic criteria and competing syndromes.

However, structured interviews themselves are not infallible.

The 2026 JAMA meta-analysis found only moderate overall test–retest reliability and considerable heterogeneity even when standardized diagnostic interviews were used.

Therefore:

Structure should support clinical reasoning, not replace it.

Layer 5 — Dimensional Measurement

Psychiatric diagnosis has traditionally been categorical.

The patient either has depression or does not.

Has ADHD or does not.

Has OCD or does not.

But symptoms exist on dimensions.

Two patients with the same diagnosis can have dramatically different symptom burdens.

Validated rating scales therefore add another layer of information.

Examples include:

PHQ-9 for depressive symptoms;

GAD-7 for anxiety;

ASRS for adult ADHD;

Vanderbilt scales for childhood ADHD;

Y-BOCS for obsessive-compulsive symptoms;

YMRS for manic symptoms;

MADRS or HAM-D for depressive severity.

These instruments are particularly useful for:

screening

quantification

baseline measurement

and

monitoring change over time.

They should not generally be treated as stand-alone diagnostic tests.

NICE explicitly states, for example, that ADHD should not be diagnosed solely on the basis of rating scales or observational information; diagnosis requires a broader clinical and psychosocial assessment with impairment across relevant settings.

A useful rule is:

Scales measure. Clinicians diagnose.

Layer 6 — Functional Validation

This is sometimes underestimated.

Psychiatric symptoms become clinically meaningful partly through their effects on functioning.

The clinician should therefore ask:

What has this symptom actually done to the person’s life?

Consider ADHD.

Occasional distractibility is extremely common.

The stronger diagnostic question is whether persistent symptoms have repeatedly affected:

• education
• work
• relationships
• finances
• driving
• organisation
• deadlines
• household responsibilities
• emotional regulation

Similarly, sadness itself does not necessarily establish major depression.

Clinical importance emerges from the combination of symptoms, persistence, associated features, distress and impairment.

Functional assessment therefore acts as another form of diagnostic validation.

Layer 7 — Exclude Alternative Explanations

The diagnostic question should never simply be:

“What psychiatric disorder does this patient have?”

It should also be:

“What else could explain this presentation?”

Potential alternatives include:

• thyroid disease
• anaemia
• nutritional deficiencies
• sleep disorders
• epilepsy
• neurological disorders
• medication effects
• alcohol or substance use
• hormonal disorders
• metabolic abnormalities
• traumatic brain injury
• neurodevelopmental conditions
• emerging neurocognitive disorders

Testing should be targeted rather than indiscriminate.

A healthy 22-year-old with a longstanding pattern of uncomplicated social anxiety does not require the same investigations as a 58-year-old with new behavioural change, cognitive deterioration and hallucinations.

The investigation should follow the clinical hypothesis.

Layer 8 — Objective Cognitive, Physiological and Biological Data

This is where psychiatry is changing rapidly.

Emerging tools include:

• computerised cognitive testing
• continuous performance tests
• neuropsychological testing
• actigraphy
• sleep measurements
• EEG
• quantitative EEG
• neuroimaging
• digital phenotyping
• speech analysis
• wearable data
• genetics
• electrophysiology
• machine-learning-based prediction

These technologies could provide valuable information.

But their role needs to be defined carefully.

Consider attention testing.

A continuous performance test may demonstrate deficits in sustained attention, response inhibition or reaction-time variability.

But:

Impaired attention does not automatically equal ADHD.

Attention can be affected by depression, anxiety, sleep deprivation, medication, neurological illness and numerous other factors.

Similarly:

Memory impairment does not automatically equal Alzheimer’s disease.

And:

An EEG abnormality does not automatically establish a psychiatric diagnosis.

A 2026 perspective in Translational Psychiatry argues that EEG biomarker research may be more fruitful when viewed transdiagnostically and dimensionally rather than by searching for simple disorder-specific signatures.

That is highly compatible with MDTP.

Biological measurements may ultimately be more useful for identifying dimensions, subtypes, prognosis and treatment response than for replacing conventional clinical diagnosis.

The MDTP Diagnostic Matrix

A practical psychiatric formulation could therefore examine evidence across several domains:

Domain	Evidence
Phenomenology	Does the symptom pattern fit?
Temporal course	Does onset and progression fit?
Functional impairment	Is clinically significant impairment demonstrated?
Collateral information	Do independent sources support the history?
Structured criteria	Are DSM/ICD criteria satisfied?
Dimensional scales	Is symptom burden objectively quantified?
Differential diagnosis	Have important alternatives been considered?
Objective testing	Do relevant cognitive/biological measures support the formulation?
Longitudinal course	Does follow-up remain consistent with the diagnosis?

Importantly, these domains should not simply be added together mathematically.

Some evidence is more diagnostically important than other evidence.

The model is intended to structure reasoning—not create an artificial score.

Diagnostic Confidence: Low, Moderate or High

The final stage of MDTP is assigning diagnostic confidence.

Low Confidence

The syndrome is plausible, but important information is missing or major alternative explanations remain.

Example:

Adult presenting with inattention, but childhood history unavailable and significant depression and sleep deprivation are present.

The appropriate conclusion may be:

ADHD — possible; diagnostic clarification required.

Moderate Confidence

Most clinical evidence supports the diagnosis, but one or more important uncertainties remain.

Example:

Longstanding symptoms suggest ADHD and current functional impairment is clear, but childhood collateral information remains incomplete.

The diagnosis could be documented as:

Adult ADHD — probable, moderate diagnostic confidence.

High Confidence

Several independent evidence streams converge.

For example:

• characteristic phenomenology
• appropriate age of onset
• longitudinal persistence
• cross-situational impairment
• collateral confirmation
• structured criteria fulfilled
• differential diagnoses assessed
• objective testing concordant where indicated
• subsequent longitudinal course consistent

The formulation could then reasonably be:

Adult ADHD — high diagnostic confidence.

This approach communicates uncertainty without abandoning diagnostic usefulness.

The Most Important Step: Try to Disprove Your Diagnosis

One of the strongest safeguards against diagnostic error is deliberate falsification.

Once the clinician believes the patient has a particular disorder, the next question should be:

What evidence would make me abandon this diagnosis?

For ADHD:

Could the symptoms be better explained by depression, anxiety, sleep deprivation, substance use or bipolar disorder?

For bipolar disorder:

Is there convincing evidence for discrete episodes of pathological mood elevation, or are we interpreting chronic emotional instability as episodic mania?

For dementia:

Could depression, delirium, medication effects, sleep disorder or another neurological condition better explain cognitive deterioration?

This transforms diagnosis from confirmation seeking into scientific reasoning.

The MDTP Clinical Workflow

The entire process can be simplified into a practical sequence:

1. Define the presenting syndrome

↓

2. Establish detailed phenomenology

↓

3. Construct a longitudinal timeline

↓

4. Assess functional impairment

↓

5. Obtain collateral information

↓

6. Apply structured diagnostic criteria

↓

7. Quantify relevant symptom dimensions

↓

8. Generate competing diagnoses

↓

9. Exclude important medical, neurological and substance-related causes

↓

10. Add targeted cognitive or biological measurements where clinically useful

↓

11. Assign diagnostic confidence

↓

12. Reassess longitudinally

This final step is critical.

Diagnosis Should Be Allowed to Change

A psychiatric diagnosis made during the first consultation should not become permanent simply because it was written first.

New information may appear.

A family member may describe previously unrecognised manic episodes.

School records may contradict the recollection of childhood ADHD.

Cognitive impairment may progress.

Substance use may become apparent.

Sleep treatment may eliminate apparent attentional symptoms.

The patient’s illness itself may evolve.

Therefore:

Longitudinal reassessment is not evidence that the original psychiatrist failed. It is part of good psychiatric diagnosis.

The purpose of follow-up is not merely to adjust medication.

Follow-up also tests the diagnostic hypothesis.

What MDTP Is Not

The framework is not an argument for ordering every available psychiatric test.

More investigations do not automatically produce more accurate diagnosis.

Unnecessary tests can create false positives and greater confusion.

MDTP instead asks:

What additional piece of information would meaningfully change my diagnostic probability?

If the answer is “none”, the test may not be necessary.

This is essentially Bayesian clinical reasoning applied to psychiatry.

Where Could Artificial Intelligence Fit?

Artificial intelligence may eventually become particularly useful in the integration stage.

A future psychiatric assessment could potentially combine:

clinical history;

structured symptom ratings;

longitudinal electronic records;

medication response;

cognitive testing;

sleep information;

wearable data;

neurophysiology;

and relevant laboratory findings.

AI may help detect patterns across these datasets that are difficult for a human clinician to process simultaneously.

But the objective should not be:

AI decides the diagnosis.

A more realistic model is:

AI organises evidence, identifies inconsistencies, estimates probabilities and assists the psychiatrist’s clinical reasoning.

Human clinical interpretation remains crucial because psychiatric symptoms exist within personal, developmental, social and cultural contexts.

From Subjective Psychiatry to Evidence-Integrated Psychiatry

The longstanding debate between “clinical psychiatry” and “biological psychiatry” may eventually prove unnecessary.

The most useful future model probably contains both.

Psychiatric diagnosis can remain clinically grounded while becoming increasingly:

structured

quantitative

multimodal

longitudinal

technology-assisted

and

explicit about uncertainty.

The question is not whether a questionnaire, cognitive test, QEEG, imaging technique or algorithm will replace clinical assessment.

The question is:

How can each source of information reduce uncertainty left by the others?

That is the central idea behind Multimodal Diagnostic Triangulation in Psychiatry.

The Future: Precision Psychiatry Begins With Better Diagnosis

Precision psychiatry is frequently discussed in terms of choosing the right treatment.

But precision treatment is impossible without sufficiently precise diagnosis and characterisation.

Two people who meet criteria for “depression” may have very different biological vulnerabilities, cognitive profiles, sleep abnormalities, developmental histories, comorbidities and treatment-response patterns.

Similarly, two patients diagnosed with ADHD may have very different patterns of attention, inhibition, emotional regulation and executive functioning.

The future may therefore move beyond asking simply:

“Which disorder does this patient have?”

towards asking:

“What is this patient’s clinical phenotype, cognitive profile, biological context, functional impairment and likely treatment-response profile?”

That represents a much larger transformation.

And it begins not by abandoning traditional psychiatric assessment, but by strengthening it with additional sources of evidence.

About Dr Srinivas Rajkumar T

Dr Srinivas Rajkumar T is a psychiatrist with postgraduate training from AIIMS New Delhi and currently works as a Senior Consultant Psychiatrist with Apollo Hospitals, Chennai, alongside his academic work in psychiatry.

His clinical interests include adult ADHD, cognitive assessment, precision psychiatry, neuropsychiatry and the responsible integration of emerging technologies into psychiatric practice.

Through ATTN Clinic, Chennai, his approach emphasises that the clinical psychiatric interview remains the foundation of diagnosis, while structured questionnaires, cognitive assessment and selected objective technologies can provide additional layers of information where clinically appropriate.

The aim is not to replace psychiatric judgement with technology, but to move towards a more structured, measurable and evidence-integrated model of psychiatric assessment.

ATTN Clinic — Attention. Understood.

Currently functioning from Apollo Clinic, opposite Phoenix Market City, Velachery, Chennai.

Appointments: +91 85951 55808

Email: srinivasaiims@gmail.com

This article discusses a proposed conceptual framework for improving psychiatric diagnostic reasoning. “Multimodal Diagnostic Triangulation in Psychiatry (MDTP)” as described here has not been independently validated as a diagnostic instrument and should not be presented or used as a replacement for established diagnostic guidelines or specialist clinical judgement.