Section 4: Digital Therapeutics: What's Ready for Clinical Use

Your patients are already using digital mental health tools. A 2023 survey estimated that more than 20,000 mental health apps were available for download, the vast majority with no clinical evidence behind them. The question is not whether digital therapeutics will enter your practice. It is whether you can distinguish the tools with legitimate evidence from the ones with polished marketing.

This section evaluates the evidence base for the major categories of technology-based anxiety interventions: internet-delivered CBT, virtual reality exposure therapy, AI-driven tools, biofeedback and neurofeedback, wearable monitoring devices, and the FDA regulatory framework that governs (or fails to govern) each.

4.1 Internet-Delivered CBT (iCBT)

Internet-delivered CBT has the strongest evidence base of any digital intervention for anxiety disorders, and for some populations, it may represent the most impactful development in anxiety treatment access since the SSRI.

The core evidence: multiple systematic reviews and meta-analyses have demonstrated that guided iCBT (where a therapist provides brief weekly feedback, typically 10 to 15 minutes per week, alongside a structured online CBT program) produces large effect sizes compared to waitlist or no-treatment controls for generalized anxiety disorder, social anxiety disorder, and panic disorder. Meta-analytic reviews by Andersson and colleagues have found pooled effect sizes in the range of d = 0.80 to 1.0 for guided iCBT versus inactive controls. Separately, non-inferiority studies comparing guided iCBT directly to face-to-face CBT have generally found comparable outcomes, though these head-to-head comparisons are fewer in number and methodologically more demanding.

The "guided" distinction matters enormously. Unguided (purely self-directed) iCBT programs show lower effect sizes and, critically, much higher dropout rates. The therapeutic relationship, even when reduced to brief weekly check-ins by text or email, appears to be a necessary ingredient for most patients to complete the program and achieve clinical benefit. This finding has direct implications for how iCBT should be implemented: it is not a replacement for human involvement but a model that extends one therapist's capacity across many more patients. (Andersson et al., 2014, World Psychiatry)

Where iCBT works best. The evidence is strongest for adult GAD, social anxiety disorder, and panic disorder with or without agoraphobia. Specific platforms with published clinical trial evidence include This Way Up (developed at St. Vincent's Hospital and University of New South Wales), MindSpot (Macquarie University), and SilverCloud (now Amwell); in the UK, NICE-recommended programs are delivered through the IAPT stepped care system described in Section 1.5. The evidence for pediatric anxiety, while growing, is less robust, and developmental adaptations (parental involvement, age-appropriate interfaces) remain underdeveloped in most available platforms. The evidence is also weaker for patients with severe anxiety, multiple comorbidities, or active suicidal ideation, who typically require higher-intensity intervention.

The scalability argument. In the context of the therapy desert described in Section 1, iCBT represents one of the few evidence-based solutions that can meaningfully increase treatment access without requiring a proportional increase in the trained workforce. A therapist providing guided support for an iCBT program can manage more patients simultaneously than a traditional face-to-face caseload, because the structured online program delivers the core therapeutic content while the therapist provides targeted feedback and support. (Andrews et al., 2018, Journal of Anxiety Disorders)

The equity question. Does iCBT reduce or widen the treatment gap? The evidence is mixed. On one hand, iCBT eliminates geographic barriers (a patient in a rural therapy desert can access the same program as one in an urban center), reduces some cost barriers, and accommodates scheduling flexibility. On the other hand, it requires reliable internet access, digital literacy, a private space for completing sessions, and a reading level sufficient to engage with text-heavy CBT materials. These requirements may exclude precisely the populations with the greatest unmet need: lower-income individuals, older adults with limited technology experience, patients with limited English proficiency, and those without stable housing. (Lattie et al., 2019, Journal of Medical Internet Research)

4.2 Virtual Reality Exposure Therapy (VRET)

Virtual reality exposure therapy addresses a persistent implementation barriers for the gold standard anxiety treatment: the logistical difficulty of conducting in vivo exposure. A clinician treating a patient with a flying phobia cannot easily arrange repeated flights. A school psychologist working with a socially anxious teenager cannot reliably reproduce the social situations that trigger the fear. Virtual reality solves these problems by generating immersive, controllable environments where the intensity of exposure can be adjusted dynamically in real time.

The evidence base varies by disorder. For specific phobias (heights, flying, spiders, enclosed spaces), VRET has the strongest support, with multiple RCTs demonstrating effect sizes comparable to in vivo exposure. (Carl et al., 2019, Journal of Anxiety Disorders) For social anxiety disorder, the evidence is growing but more limited, with the challenge being that social scenarios require more complex and realistic virtual interactions than phobia-related stimuli. For agoraphobia and panic disorder, preliminary evidence is promising but the trial base is small.

A practical consideration: the cost and accessibility of VR equipment have decreased in recent years, making clinic-based VRET feasible for many practice settings. The primary limitation is no longer the hardware but the availability of clinically validated software programs; many commercially available "anxiety VR" applications have not undergone rigorous clinical testing. (Carl et al., 2019, Journal of Anxiety Disorders)

4.3 AI and App-Based Tools: Evidence Versus Marketing

The gap between marketing claims and clinical evidence is wider in the mental health app space than in almost any other area of healthcare technology.

Tools with RCT evidence. A small number of AI-driven conversational agents have been evaluated in randomized controlled trials. Woebot, a chatbot delivering CBT-based techniques, has published RCT data showing modest reductions in anxiety and depression symptoms compared to information-only control conditions, with the strongest evidence in college student populations. Wysa has demonstrated similar preliminary efficacy. (Fitzpatrick et al., 2017, JMIR Mental Health)

The important caveat: these trials typically compare the app against an inactive control (information provision or waitlist), not against active treatment. Demonstrating that a chatbot is better than nothing is a different claim than demonstrating it is a viable substitute for therapist-delivered CBT. The effect sizes reported in app studies are generally smaller than those for guided iCBT or face-to-face therapy.

The vast majority of mental health apps available on commercial app stores have no published clinical evidence whatsoever. Systematic reviews have found that fewer than 5% of mental health apps on major platforms have any peer-reviewed evidence supporting their claims.

The American Psychiatric Association's App Evaluation Model

provides a useful clinical framework: digital tools are best positioned as adjuncts to clinical care (e.g., between-session practice tools, mood tracking, psychoeducation delivery) rather than standalone replacements for evidence-based treatment. For the small number of tools with published RCT data (Woebot, Wysa), cautious recommendation as an adjunct or bridge, particularly for patients on waitlists or in areas without access to guided iCBT, is clinically appropriate. For patients using or considering tools without published evidence, clinicians should counsel that the evidence base is limited and encourage engagement with evidence-supported platforms over commercial wellness apps. (American Psychiatric Association, App Evaluation Model)

4.4 Biofeedback, Neurofeedback, and Wearables

Biofeedback (training patients to modulate physiological signals such as heart rate variability, respiratory rate, or electrodermal activity) has a decades-long research history in anxiety. The evidence is modest: meta-analyses consistently find small to moderate effect sizes, generally inferior to CBT. Biofeedback may have a niche role as an adjunct for patients who are resistant to cognitive approaches or who benefit from a concrete, physiological demonstration that they can influence their own arousal. (Goessl et al., 2017, Psychological Medicine)

Neurofeedback (training patients to modulate EEG patterns, typically targeting increased alpha power or reduced frontal asymmetry) has a weaker evidence base. Trials are generally small, methodologically variable, and plagued by inadequate sham controls. Current evidence does not support neurofeedback as a standalone anxiety treatment. (Begemann et al., 2016, Translational Brain Rhythmicity)

Wearable monitoring devices (smartwatches, wristbands) that track heart rate variability, electrodermal activity, and sleep patterns are increasingly marketed for anxiety management. The technology for detecting physiological correlates of anxiety states is advancing rapidly. The clinical problem is translation: knowing that a patient's HRV decreased at 2:00 PM on Tuesday does not, by itself, generate a clinical intervention. The most promising use case is passive monitoring integrated with just-in-time adaptive interventions (JITAIs), systems that detect physiological markers of rising anxiety and deliver a brief intervention (breathing exercise, grounding technique) in real time. This approach is in early-stage research and not yet clinically validated for anxiety disorders. (Jacobson et al., 2019, npj Digital Medicine)

Digital Therapeutics for Anxiety Disorders, Evidence and Readiness Summary

Modality	Evidence Quality	Best Supported Disorders	Key Practical Barriers	Regulatory Status
Guided iCBT (e.g., This Way Up, MindSpot, SilverCloud)	Strong (multiple MAs, non-inferiority data)	Adult GAD, social anxiety, panic	Requires internet, digital literacy, reading level; limited pediatric evidence	Not FDA-regulated as devices; delivered as healthcare services
Unguided iCBT	Moderate (lower effect sizes, high dropout)	GAD, social anxiety	No therapist support increases attrition; effects smaller than guided	Same as above
VR exposure	Moderate (multiple RCTs for specific phobias)	Specific phobias (heights, flying, spiders)	Validated software limited; social anxiety scenarios less developed	No therapy anxiety-specific FDA clearance
Woebot (AI chatbot)	Preliminary (RCTs vs. inactive controls)	Depression, anxiety in college students	Not compared to active treatment; commercial product	Not FDA-cleared for anxiety
Wysa (AI chatbot)	Preliminary (RCTs vs. inactive controls)	Anxiety, depression	Similar limitations to Woebot	Not FDA-cleared for anxiety
Biofeedback (HRV, respiratory)	Modest (small-moderate effect sizes, inferior to CBT)	Niche adjunctive role	Requires equipment, training; not standalone treatment	Variable; some devices FDA-cleared as biofeedback, not as anxiety treatment
Neurofeedback (EEG)	Weak (small trials, poor sham controls)	Not established	Equipment cost, expertise required; evidence insufficient	Not FDA-cleared for anxiety
Wearable monitoring + JITAI	Very early (proof-of-concept)	Not established	Translation from detection to intervention unvalidated	Not FDA-cleared

Note: "Evidence Quality" reflects published clinical trial data for anxiety disorders specifically as of early 2026. Most digital tools have more evidence for depression than anxiety. "Regulatory Status" refers to the US FDA regulatory landscape; other jurisdictions may differ.

4.5 The FDA Regulatory Landscape

The regulatory framework for digital therapeutics is evolving and clinicians should understand what FDA designations do and do not mean.

Prescription Digital Therapeutics (PDTs) are software-based interventions that have undergone FDA review and are intended to be prescribed by clinicians. The FDA has created specific regulatory pathways for these products, including the De Novo classification and the Digital Health Center of Excellence. Somryst for insomnia (Pear Therapeutics, FDA authorized 2020) was one of the first PDTs to receive FDA authorization, and SleepioRx (Big Health, FDA cleared 2024) subsequently demonstrated the same principle: that software-based interventions can meet the evidence standards traditionally applied to drugs and devices. (FDA Digital Health Center of Excellence)

For anxiety specifically, no PDT has yet received full FDA authorization as a standalone treatment. Several are in the pipeline. The important distinction for clinicians: FDA "clearance" (the 510(k) pathway, demonstrating substantial equivalence to an existing device) is a lower bar than FDA "approval" (the PMA pathway, requiring clinical trial evidence of safety and efficacy). Many digital health products marketed for anxiety have received neither. Terms like "FDA-registered" or "built on evidence-based principles" that appear in marketing materials do not indicate that the product has been evaluated by the FDA for clinical effectiveness.

Clinical Pearl: Talking to Patients About Mental Health Apps
When a patient asks about a mental health app, three questions can structure the conversation. First: has this specific app been evaluated in a published clinical trial? (Not "is CBT effective?" but "has this app's delivery of CBT been tested?") Second: what is the comparator? An app that outperforms a waitlist condition is not the same as one that matches or approaches face-to-face therapy.
Third: what happens to your data? Many commercial mental health apps collect sensitive information (mood logs, symptom reports, location data) and their privacy practices vary widely.
The APA recommends that clinicians help patients evaluate apps using established frameworks like the APA's App Evaluation Model, which assesses accessibility, privacy, clinical evidence, interoperability, and engagement. In plain language for families: *Technology is creating new ways to access anxiety treatment, but not all digital tools are created equal. Here is a quick guide to what the evidence supports.
Internet-based CBT programs, where you work through structured lessons online with brief weekly check-ins from a therapist, have the strongest evidence: for some anxiety disorders, they work about as well as meeting with a therapist in person. Virtual reality therapy for specific fears (heights, flying, spiders) also has good research support.
Mental health apps and AI chatbots are more of a mixed picture: a very small number have been tested in clinical trials and show modest benefits, but the vast majority of the thousands of apps available have never been studied at all. The label "evidence-based" on an app store listing does not mean the app itself has been tested. It usually means the app uses general principles from therapy that has been tested, which is a very different claim.
If your family member is considering using a digital tool for anxiety, the most important question is whether it has published research supporting that specific product, not just the general approach it claims to use.*

PsychHQ

Digital Therapeutics