Over the years I have seen this many times, as has any physician who has practiced long enough. Two patients come in with the same diagnosis, labs, pathology report, and treatment plan, whether managing a complex melanoma or a chronic inflammatory disease. Yet their bodies respond differently.

We spend our training learning to explain the disease and the drug, and build almost nothing around explaining why individuals respond so differently to the exact same therapy. That mystery is one of the oldest in medicine, and now a second pressure sits on top of it. As artificial intelligence gets better at the clinical tasks we spent decades training for, we are forced to ask: What is left that only a physician can do? Reading a scan, reviewing a pathology report, matching a pattern, choosing among drugs: These are tasks AI is built for, and will only improve.

But I do not find this threatening; I find it clarifying, and empowering. The more the machine takes over, the harder it becomes to avoid asking what truly remains of the practice of medicine. The honest answer sits in the space between those two patients, in everything we decided we could not easily measure and so quietly dismissed as “not really medicine.”

The body processes healing through three distinct languages. These three are not equally understood. We are fluent in the first, still learning the second, and nearly illiterate in the third. But understanding is not the same as importance. In their effect on whether a patient heals, the three may matter far more equally than the attention we have paid them. We have mastered the language that was easiest to measure.

1. The language of pharmacology

This is biochemistry. A molecule binds a receptor, a pathway shifts, and the intended effect follows. We trust it most because it is concrete, measurable, and reproducible. It is pattern work, and pattern work is what machines do best. Though vital, it is only one of three. Fluency in one language is not fluency in medicine.

2. The language of biophysics

This is the language of force and structure. Tissue is not a passive bag that chemistry happens inside. Cells sense physical pressure, tension, and stiffness, and respond. We see it whenever the heart thickens under the chronic pressure of hypertension, or an artery remodels its wall to the force of the blood inside it: structure rewritten by force, not by chemistry. The body’s physical state is a dynamic mechanical system, and machines are only beginning to model it.

3. The language of psychoneuroendocrineimmunology (PNEI)

This language is the one most of us were trained to call “soft,” unmeasurable, or not quite credible. It is also the one a machine cannot take over, because we have barely learned to quantify it. PNEI studies how perception, stress, and emotion translate into systemic biology through the nervous, endocrine, and immune systems acting as one communication loop. Chronic stress raises cortisol; vagal tone, readable in heart rate variability, tracks directly with these inflammatory markers. The mind does not hover above the body; it is wired into its chemistry, a live current.

We have mastered the first language and begun on the second, but the third has been left largely ignored. How many CME courses are devoted to it? We are rarely taught to activate this terrain in our patients, to turn them from passive recipients into active participants. Instead, we have dismissed it as bedside manner and ignored its therapeutic potential.

AI cannot create trust, meaning, safety, or the felt experience of being cared for. That remains the physician’s work. But what a machine cannot generate is not a reason to lock it out of the exam room. We assume technology pulls us away from the human core of medicine. Used deliberately, it may instead carry that work further than a rushed clinical visit ever could.

Which brings me to the word physicians least like to say in a clinical sentence: spiritual. I am not claiming belief transcends biology, or substitutes for real treatment. The point is to recognize that meaning, fear, hope, and trust enter the body through physiology. The safety a patient carries out of the clinic manifests in stress hormones, in inflammation, and in the physiological terrain the rest of care must work in. Understood this way, the human side of healing is one of the body’s own core mechanisms. We confounded “we haven’t learned to measure this yet” with “this isn’t real.”

This changes the care model in two ways. The felt sense of meaning and safety alters real physiology. And a patient who understands their illness becomes a motivated agent in their own recovery, changing adherence, not the brittle kind that comes from being told what to do, but the durable kind that comes from conviction. A drug only works when it is taken; an informed patient works continuously. We have a drug for the receptor and almost nothing for the person who carries it home.

I have come to call this the engineering of healing, and PNEI is its physiological infrastructure. The body heals itself; it always has. Every moment, its biochemistry and immune system repair and restore it. This is the physician’s highest work: to learn all three languages and orchestrate them at once, the molecular, the physical, and the human, so the body’s own healing runs as far as it can. Each, followed far enough, is physical.

None of this replaces evidence-based medicine. But the same machines that are absorbing the measurable parts of medicine are, by stripping them away, revealing the part we have always overlooked. Naming it is most of the work. What we have called soft, unscientific, or merely human turns out to be physiology we never learned to read, and recognizing that is where the change begins.

Healing, in the end, is physical all the way down, and human connection is part of that physics.

Michael P. Tabibian is a dermatologist.