The Future of Precision Psychiatry
Exploring new technologies that provide personalized mental health treatment.
Imagine a version of mental health care that adapts to you — your brain, your body, your life — as you go. A personalized plan that can adjust in real time, combining brain stimulation, psychotherapy, medications when useful, and lifestyle rhythms. That’s the future of precision interventional psychiatry — and pieces of it are already here. If and when we arrive at the “AI Singularity” that Kurzweil envisions, that could suddenly level up beyond what we can currently imagine. The uncertainty is almost unbearable.
If we can fulfill this promise, mental illness, as well as personal enhancement, could make the proverbial jump to light speed.
The State of the Art Is a Moving Target
We can currently personalize where we stimulate the brain for many conditions, such as clinical depression. Specific networks matter enormously, such as the default mode network (DMN), which appears core to consciousness. Stimulating the right spot or spots on the surface of the brain’s cortex can influence these mood networks, like the DMN, and the Fronto-Paretal and Salience Networks, more effectively, either directly or indirectly. We can do this now, with transcranial magnetic stimulation (TMS)1. The key is that the “right spots” — like the connection between prefrontal control areas and deeper mood circuits — differ for each person by a centimeter or two, which is huge when you’re using a focused magnetic field. While standard approaches work well and may be appropriate in many cases, at least for some, neuronavigation makes a significant difference. And TMS already vastly outperforms multiple medication trials. Dynamic causal modeling, digital twinning, and other computational and related artificial intelligence/machine learning (AIML) approaches provide deep understanding of what actually causes improvement or treatment failure.
We can monitor and model the brain in real time, as well as with static imaging. During noninvasive brain stimulation with TMS, tools like electroencephalography (EEG) or functional near-infrared spectroscopy (fNIRS2) can show the brain response with millisecond resolution, or good spatial resolution, respectively. We can adapt on the fly to changing conditions to optimize treatment parameters like timing, dose, and location.
Why fNIRS? EEG offers great temporal resolution, but spatially it is coarse-grained: shine harmless near-infrared light into the scalp and measure tiny changes in oxygenated and deoxygenated blood in the cortex, and we can get an anatomically precise map. Because active brain regions draw more oxygenated blood, fNIRS can infer when parts of the outer brain are working harder during tasks or rest. When these are used together, we could get office-ready spatial and temporal resolution for real-time treatment planning, prediction, response optimization, and post-treatment monitoring for recurrence prevention and response maintenance.
A New Vision for Mental Health
Traditional treatment is limited, involving trial-and-error and time lags in treatment response and monitoring. For example, it takes many weeks of antidepressant treatment to even know if the treatment is working, based on clinical interviews. Wouldn’t it be great to use something like fNIRS to see in a week if therapy or medications are working, or even to help select the best options to inform treatment planning? Techniques like pharmacogenomics, which are meant to help with medication selection by looking at patient genetics, have fallen short of their promise because mental illness has so many genetic influences. Personalized interventional psychiatry aims to shorten that path by matching the intervention to your brain and adapting as we go.
Here’s how a course of care might look for someone with persistent depression:
We begin with a brain MRI to map your networks and identify your personal stimulation target — often in the left dorsolateral prefrontal cortex, where it connects inversely to a deep mood area called the subgenual cingulate. We use TMS-EEG/fNIRS to find the coil position and orientation that gives a clean, strong early brain response. We might treat more than one region and then adjust where we treat based on monitoring during treatment. We use similar tools to see if the treatment is having the desired effect, and likewise afterward to see if it is durable, or if we need to provide additional treatment.
This combines with similar input regarding therapy and lifestyle. Wearables now give simple measures of sleep, heart rate variability, and stress patterns, tracking actigraphy and heart rate variability, using natural language processing to pick up on communication shifts, and so on. Brief daily check-ins on mood and energy tell us how you’re doing between visits. AI can help organize these signals into a personal model and provide input to human clinicians to use to augment patient care. Privacy measures are put in place to ensure confidentiality and proper data use.
As symptoms improve, we protect the gains — spacing booster sessions based on how your system holds up, scheduling brief therapy tune-ups during stress, and using simple early-warning signs to get ahead of relapse. The goal isn’t just less depression, it’s more resilience, and over the longer haul, to “bend the developmental curve” using TMS-assisted psychotherapy (TAP), or some version thereof.
Looking Ahead
The human touch, I believe, will always matter and can never be replaced. Feeling safe, seen, and engaged in meaningful work will always be therapeutic and essential. Feelings aren’t noise to be cancelled; they’re signals that tell us what matters and whether we’re overdoing it or not doing enough. Patients are experts on their own experience, and there is no substitute for clinician intuition and judgment, though when machine learning allows for superior outcomes, humans must listen. This is where human and machine models complement each other: Psychotherapy helps you make meaning and build skills; brain stimulation can help by both increasing neural flexibility (for example, in autism or OCD) and entraining patterns that work better.
More clinics will offer individualized targeting as a standard, not an extra, though there are barriers such as cost, coverage, and ease of use. We need to know when to use more advanced techniques and when simpler approaches suffice. People will feel more like collaborators than test subjects — perhaps with the aid of AI-augmented tools, or even assistants to bridge the gaps in between meetings — seeing their own maps in a way we can understand and use, gaining insight into choices in a way that supports motivation and change, and using wise and intelligent emerging approaches to raise the bar on what we can expect in mental health care.
At the same time, there are many precautionary notes. Neuroscience is beginning to pick up, after decades of lagging behind predictions. Developments in AIML and computational psychiatry will help drive that forward. But the more we know about ourselves, and the more that information is available for public scrutiny, the greater the risk of misuse of that same information, for example, to deny insurance claims or coverage for those in higher-risk groups, or for commercial use such as in marketing and advertising. Likewise, AI hype is a problem, but there is a fair chance that the more ambitious predictions will come to fruition faster than we realize. So we must prepare for that eventuality with safer AI mental health guardrails, as we may see artificial general intelligence, or even artificial superintelligence, sooner than we expect.
References
- I’ve been treating patients with TMS since 2010, and in that time, especially the last few years, techniques like accelerated TMS and MRI-guided TMS have advanced tremendously. I’ve also treated lots of folks without using neuronavigation, rather standardized localization protocols, who have had excellent and life-changing responses. Though not all do, and maybe those folks would do better with navigation, if it were more readily available. We do offer neuronavigation, but insurance doesn’t yet cover it.
2. fNIRS is one of many emerging technologies. Others (advanced EEG, digital phenotyping, novel forms of neuromodulation) also show promise. The art is choosing what adds value for each person, keeping the focus on outcomes and experience — not on gadgets.
Băcilă, C. I., Marcu, G. M., Vintilă, B. I., et al. (2025). Applications of functional near-infrared spectroscopy (fNIRS) in monitoring treatment response in psychiatry: A scoping review. Journal of Clinical Medicine, 14(15), 5197. https://doi.org/10.3390/jcm14155197
Chen, W.-L., Wagner, J., Heugel, N., et al. (2020). Functional near-infrared spectroscopy and its clinical application in the field of neuroscience: Advances and future directions. Frontiers in Neuroscience, 14, 724. https://doi.org/10.3389/fnins.2020.00724
Devezas, M. Â. M. (2021). Shedding light on neuroscience: Two decades of functional near-infrared spectroscopy applications and advances from a bibliometric perspective. Journal of Neuroimaging, 31(4), 641–655. https://doi.org/10.1111/jon.12877
Ehlis, A.-C., Schneider, S., Dresler, T., & Fallgatter, A. J. (2014). Application of functional near-infrared spectroscopy in psychiatry. NeuroImage, 85, 478–488. https://doi.org/10.1016/j.neuroimage.2013.03.067
Ho, C. S. H., Lim, L. J. H., Lim, A. Q., et al. (2020). Diagnostic and predictive applications of functional near-infrared spectroscopy for major depressive disorder: A systematic review. Frontiers in Psychiatry, 11, 378. https://doi.org/10.3389/fpsyt.2020.00378
Kurzweil, R. (2024). The singularity is nearer: When we merge with AI. Viking.
Yeung, M. K., & Lin, J. (2021). Probing depression, schizophrenia, and other psychiatric disorders using fNIRS and the verbal fluency test: A systematic review and meta-analysis. Journal of Psychiatric Research, 140, 416–435. https://doi.org/10.1016/j.jpsychires.2021.06.015
This Blog Post (“Our Blog Post”) is not intended to be a substitute for professional advice. We will not be liable for any loss or damage caused by your reliance on information obtained through Our Blog Post. Please seek the advice of professionals, as appropriate, regarding the evaluation of any specific information, opinion, advice, or other content. We are not responsible and will not be held liable for third party comments on Our Blog Post. Any user comment on Our Blog Post that in our sole discretion restricts or inhibits any other user from using or enjoying Our Blog Post is prohibited and may be reported to Sussex Publishers/Psychology Today/Medium/SubStack. Grant H. Brenner. All rights reserved.
Originally published on Psychology Today ExperiMentations
