For at least the last decade, digital mental health technologies (DMHTs) such as smartphone apps, virtual reality, and wearables have been expanding in scope and potential. Today, generative artificial intelligence (AI) has joined the list of guided (coached) and unguided (self-help) digital tools that aim to deliver mental health interventions. A groundswell of interest and investment in these technologies underscores their potential to increase access to care and deliver scalable interventions. And yet the clinical benefits of these DMHTs remain largely unrealized. How can research, investment, and innovation better align to improve mental health outcomes for patients? The first step is to move past the appealing narrative of a dichotomy between wellness and health devices. Today, wellness DMHTs such as mindfulness apps are often viewed as having no risks and being exempt from requiring substantial data to prove efficacy. Medical DMHTs, such as prescription therapeutics, present the opposite scenario, often claiming substantial potential risk and presenting (often weak) efficacy data. As the field faces the next challenge of engagement, transparency in safety and efficacy rather than categorical wellness or medical labels will be critical to ensure future innovations are better able to engender trust, deliver benefits, and catalyze engagement.

In the wellness space, DMHTs are perceived to carry no risk because they are akin to wellness products. But nearly a decade in, evidence for harm from even these wellness-focused treatment tools is mounting. Although few studies thoroughly report on harm today, as large-scale digital health studies become available they reveal surprising results. In a 2022 study of nearly 19鈥�000 people with frequent suicidal ideation, those randomized to an online low-intensity format of dialectical behavior therapy had higher rates of nonfatal or fatal self-harm over the 18-month trial compared with control groups.¹ A lack of full awareness of potential harm may have led to the 2023 rollout of the Tessa chatbot for eating disorders by the National Eating Disorders Association without a clear safety plan鈥攔esulting in users taking to social media to beg that the chatbot be turned off before the association realized the full extent of harm. Privacy concerns have also become rampant, with the Federal Communications Commission calling for a probe into how the Crisis Text Line service shared users鈥� text messages with a sister for-profit AI company in 2022. Less visible harms (eg, inequities in digital access and digital literacy) exclude vulnerable populations from receiving potential benefits. Although the benefits of a wellness approach can ensure products reach users faster and are not slowed by regulation, the rate of dangerous products and concerning privacy practices appears to be accelerating. Instead of assuming no risk, we must discuss implementing safety plans offered by DMHTs that go beyond asking users to call helplines. We need to start reporting harm so that we can create plans to minimize risk. Labeling a digital health treatment as wellness does not abnegate the need to ensure products are safe, private, and accessible to all.

For those digital mental health interventions that have taken a more medical approach, the focus is often on proving their product is evidence based. However, the current quality of that evidence is often limited even for those DMHTs that have attempted to seek formal US Food and Drug Administration (FDA) clearance as medical devices. An analysis of clinical studies conducted on digital health products authorized by the FDA as of November 2022 found the majority lacked rigor, with most studies conducted on a postmarket basis, without blinding or a digital control group, and self-funded by manufacturers.² In 2022, the Centers for Medicare and Medicaid Services announced they would not provide automatic Medicare coverage, noting FDA-cleared devices often lacked sufficient protection for patients and evidence of clinical benefit.³ Concerns about research quality apply broadly, with text-messaging interventions also criticized for having lower-quality research and inconclusive results regarding benefits in depression or anxiety.⁴ Nonetheless, in November 2023, New York City signed a $26 million contract with a text-messaging鈥揵ased company to increase access to mental health services for children, highlighting that clinical efficacy is likely not yet the prime driver in business or use decisions. Without stronger clinical evidence, the current DMHT ecosystem appears to compete more on marketing than scientific evidence. The dangers of relying on this approach are highlighted in a 2022 review of digital health start-ups that found no correlation between their clinical robustness and either the number of clinical claims or total funding raised.⁵ Now is the time to demand evidence and look beyond marketing to ensure the quality of research matches the highest scientific standards. Systematic scientific work will likely yield more rapid progress than the current fragmented product-focused ecosystem.

Although wellness products are not inherently safe and medical ones not inherently efficacious, a common challenge unites them: engagement. During the height of the COVID-19 pandemic, Reno, Nevada, spent $1.3 million for a text-messaging鈥揵ased offering for its residents but opted not to renew the contract, with reports that only 0.5% of residents engaged with the service. This engagement challenge is ubiquitous in DMHTs and has been well known for at least the last 20 years.⁶ With the lack of success of any digital means鈥攅ven gamification鈥攖o transform engagement, there has been a recent turn to coaches, often called digital navigators,⁷ to support technology use. The effects can be striking. In one study, an app that included human support was reported to be effective for both depression and anxiety,⁸ but when offered during the height of the COVID-19 pandemic to approximately 50鈥�000 college students without human support and study incentives, only 117 downloaded that same app.⁹ These engagement challenges remain poorly understood, with a 2022 systematic review and meta-analysis of nonclinician guidance on the effectiveness of digital mental health interventions suggesting the potential of human support¹⁰ and formalized digital mental health intervention coaches in the role of digital navigators gaining traction.⁷

The current dichotomy between wellness and medical DMHTs for treatment has likely resulted in harm to patients, uncertainty about impact, and a lack of scientific progress. Focusing less on this dichotomy and instead on safety, evidence, and engagement will catalyze the next phase of DMHTs. Not all DMHTs need to move away from wellness or involve FDA oversight (tools could argue for 鈥渆nforcement discretion鈥� based on evidence of low risk). Toward the goal, we provide the following suggestions:

• Patients and clinicians should not assume wellness digital health technologies are always dangerous, nor should they assume health technologies are always safe. Instead, they should demand rigorous clinical study data (with digital control conditions) in addition to clinical use data.

• Payers should compensate digital health technologies with risk-based contracts based on engagement and clinical outcomes instead of fixed-price contracts based on access to the technology.

• Regulators should better enforce the delineation of health and wellness products and discourage the current trend of products skating the line between both.

• Researchers should focus on replicable and mechanistic outcomes for digital health technologies instead of focusing on individual products.

• All parties should support and encourage digital equity through concomitant efforts to support digital access, digital literacy, and digital navigators so that everyone has the chance to benefit.

Enabling competition based on safety, outcomes, and engagement will help guide and optimize new research investments in DMHTs, whether they are reported to be wellness or medical focused. The potential that has sparked global interest in DMHTs remains but now requires high-quality research and clinical data to prove their value.

Corresponding Author: John Torous, MD, Department of Psychiatry, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Rabb-2, BIDMC, Boston, MA 02446 (jtorous@bidmc.harvard.edu).

Published Online: April 17, 2024. doi:10.1001/jamapsychiatry.2024.0532

Conflict of Interest Disclosures: Dr Torous reported being a member of the Scientific Board for Precision Mental Wellness outside the submitted work. Dr Firth reported being supported by a UK Research and Innovation Future Leaders Fellowship and having received honoraria and consultancy fees from Atheneum, Informa, Gillian Kenny Associates, Bayer, Big Health, Hedonia, Strive Coaching, Wood For Trees, Nutritional Medicine Institute, Angelini, ParachuteBH, Richmond Foundation, and Nirakara independent of this work. Dr Goldberg reported receiving grants from the National Center for Complementary and Integrative Health and the Hope for Depression Research Foundation during the writing of this article. No other disclosures were reported.