How Modern Technology Can Individualize and Scale Depression Treatment

Here are three patients with depression:

• Patient A fears antidepressant medication and chooses, rather, to cope through excessive alcohol drinking.
• Patient B is diagnosed with both depression and borderline personality disorder, being treated with psychodynamic psychotherapy, but whose symptoms only resolve upon a subsequent diagnosis of ADHD and concurrent stimulant medication.
• Patient C experiences panic attacks with a preferred depression medication; his condition improves through sleep medication and brief interpersonal therapy.

How do three people with depression have such vastly different problems and solutions?

Unlike illnesses in other fields of medicine, mental health conditions under real clinical settings rarely abide by the textbook. This variability makes conditions difficult to diagnose and solutions imprecise. However, mental illnesses have biological causes, and their treatments have a biological effect; we simply do not yet utilize the right tools to tackle them.

Our current diagnostic system is based solely on psychological phenomena: human cognition, emotion, and behavior. Psychiatric assessments require that clinical evaluators possess an in-depth understanding of such as they engage in an extensive interview process with the patient. Human interaction is the main form of testing; besides ruling out non-psychiatric conditions, laboratory tests and imaging technologies cannot point to a specific mental diagnosis.

However, over the last few decades, we have started to understand the biochemical changes in the endocrine and immune systems of patients with mental illness, it is not observed in all patients and thus insufficient to be biological markers for diagnosis. Similarly, through the advances of neuroimaging techniques and burgeoning studies, we have begun to capture subtle abnormalities in the brains of people with psychiatric conditions; but again, these abnormalities are not found in all patients and are insufficient as biological markers. Thus, clinicians must probe the affected organ—the brain—and make precise interpretations with only, ironically, their own brains in an individual patient throughout the course of multiple sessions.

How and why do current treatments for depression fall short?

Current treatments fall under two categories and work with varying efficacies. The biological treatments include medication and neurostimulation: the former targets neurotransmitters, receptors, and transport molecules, and the latter focus on electrical patterns in the brain that regulate the electrical signaling of the brain. However, a common trend still appears—there are no consistent markers that predict who will respond to which treatment (if at all). And, in large population studies, these treatments are only slightly more effective than placebo, indicating that we still lack a clear understanding of the biological underpinnings of mental illness.

Given that the heritability of depression lies somewhere between 30-40 percent, which means that less than half of the contribution to depressive illness is due to genetic (non-developmental and non-environmental) risk factors, psychological and social factors make the other half of the contribution to illness. Psychosocial treatments, which mainly constitute psychotherapy and works for some patients with certain mental illnesses like anxiety and depression. However, given the broad array of theoretical stances from which techniques emerge, it is still unclear (clinically and biologically) which psychotherapies work for who and why.

Fortunately, we do know that for anxiety and depression, psychotherapy and biological treatments together have beneficial synergistic effects. However, with each individual patient responding differently to varying combinations of the large array of psychotherapies and biological treatments, the clinician and the patient are left with too many options and not enough direction. This trial and error is unacceptable. The most tragic consequence of our failures in the mental health field is when patients suffer, unnecessarily, from the imprecise, impersonal diagnoses and treatments.

How can “Big Data” analytics create a mental health delivery system to individualize diagnosis and treatments in mental health?

Let's return to our three initial cases. As you might have intuited, though each patient had been diagnosed with depression, its presentation—from underlying psychological factors to comorbid mental illnesses, to neural pathways—varied. "Traditional" medicine failed to provide adequate outcomes, and the path to improvement was often circuitous. Thus, it is crucial that we reassess our approach to mental health as we forge ahead. Especially with the increasing prevalence of mental illnesses within our society¹ and the added impetus of this pandemic (I explore its effects on mental health in a different post²), there is an urgency for the creation of more precise diagnostic and treatment methods. Yet, we need not look far.

Never before have we been able to analyze video, audio, and written language using machine learning statistical modeling techniques. From these methods, we can obtain cognitive, behavioral, and emotionally laden information from patients, analyze and interpret dense data, and use results to both aid in diagnosis and deliver precise care to each and every patient. Although no machine or digitalized program will replace the ever vital clinician-patient relationship, we ought to rely on these technological advances for greater diagnostic accuracy and individualized techniques to refine our care in the field of mental health. This modernized evidence-based individualized approach has been too-long delayed.