The complexity and heterogeneity of mental disorders has challenged psychiatry since the inception of the field (1). Commonly used diagnostic markers, such as blood tests, radiologic findings, and electrophysiological measurements can be inadequate to diagnose mental disorders accurately, and the severity of the symptoms (2). Advances in genetics and in neuroimaging offer new tools that psychiatry have embraced to advance understanding of the genetic and neural basis of psychiatric disorders (3). Even more recently, smartphones and wearable sensors/devices have been proposed as another set of tools for advancing understanding of physiological and behavioral perspectives of these disorders over time (4). Smartphone ownership is estimated to be 83.4% (5), and a willingness has been shown by patients to use a smartphone to monitor their mental health (6, 7). Growing literature in mental health points to the potential of smartphones and other personal digital devices to increase access to care (6), improve diagnosis, and enable remote monitoring (2, 8).

Termed “context sensing,” “personal sensing,” or “mobile sensing” by computer scientists, medicine has adopted the term “digital phenotyping” (9). Digital phenotyping, defined as the moment-by-moment quantification of the individual-level human phenotype in situ using data from smartphones and other personal digital devices, holds considerable potential for psychiatry, and the collection of phone-mediated social and behavioral markers may offer a new target for biological psychiatry (10). Digital phenotyping consists of the continuous acquisition of various metrics, including an individual’s physical activity and location, voice and speech patterns, and human-machine interactions, as sociability metrics (11, 12). By providing a highly granular view of these social and behavioral variables, mobile technology has the potential to transform current medical practice, reshape behavioral sciences, and improve the ability to deliver behavioral health treatments, thus creating a more holistic view of each patient’s trajectory in terms of mental and physical health (1, 9, 12–15).

Depression affects 322 million people worldwide and, in 2017, was considered the largest contributor to disability (16). According to the 5th edition of the Diagnostic and Statistical Manual of Mental Disorder (DSM-5) diagnostic criteria, major depressive disorder (MDD) is characterized by the presence of depressed mood or anhedonia for at least 2 consecutive weeks, during which abnormalities of neuro-vegetative function (appetite, weight loss, and sleep disturbance), psychomotor activity (e.g., loss of energy, agitation, or retardation), and cognition (feelings of worthlessness or inappropriate guilt), as well as anxiety and suicidal ideation, appear. Symptoms must be present most of the day and nearly every day (17). MDD is highly prevalent, with lifetime prevalence ranging from 2 to 21% (18), and considered one of the most disabling and burdensome mental illnesses (18, 19).

Depression is associated with several behavioral components (e.g., reduction in activity, psychomotor retardation, and changes in sleep), and motivational states (e.g., anhedonia), some of which may be detectable using a mobile phone or wearable device (20, 21). Recent technology has a significant potential for innovation to monitor behavioral and environmental risk indicators and to improve the long-term management and treatment of people suffering from depression (22).

Trazodone hydrochloride is a triazolopyridine derivative that was synthesized in Angelini Pharma laboratories in the 1960s. It is the first serotonin receptor antagonist and reuptake inhibitor developed for the treatment of depression (23). Trazodone is currently approved and marketed in several countries worldwide for the treatment of MDD, with or without anxiety, in adult patients. Relating to its pharmacological actions in humans, trazodone is defined as an up-to-date multimodal (24) and multifunctional drug with dose-dependent activity (25).

The trazodone once-a-day (OAD) formulation (available in 150 and 300 mg bisectable tablets) allows, by using a unique drug-delivery technology, release of the active ingredient over 24 h to improve treatment adherence and to provide an effective antidepressant dosing (300 mg/day) through a single administration. The pharmacokinetic profile of trazodone OAD is characterized by a slow increase of plasma level with a single low and delayed peak followed by a slow decline in plasma concentration, resulting in fewer associated adverse events (AEs) like sedation or hypotension (26).

Since its approval, the antidepressant efficacy and the favorable tolerability profile of trazodone have been confirmed in several pharmacological and clinical studies (27, 28). Results have demonstrated that trazodone was as effective as other antidepressant classes in the management of depressive disorders (26, 29–35). Contraindications for the use of trazodone OAD include known hypersensitivity to trazodone and any of the excipients, alcohol intoxication and intoxication with hypnotics, and acute myocardial infarction. The most common side effects include headache, dizziness, and sedation.

The aim of the study was to describe the clinical characterization of patients affected by MDD and treated with trazodone OAD monotherapy, applying the digital phenotyping approach. The study objectives and endpoints are presented in Table 1.

This observational study aimed to explore the potential of the digital phenotyping approach in describing the clinical course and outcome in patients with MDD treated with trazodone OAD monotherapy. During this 8-week study, the clinical characterization of patients affected by MDD was described using three methods: digital phenotyping (passive data collection by smartwatch), active data collection (completion of an online survey by the patient), and traditional in-clinic data collection (completion of the MADRS by the investigator). The description of functional parameters, collected passively, and the relationship between these data with data collected actively (ePRO) and the clinical evaluation was also explored. The implementation of the digital phenotyping method, together with traditional in-clinic data collection, aimed to engage patients’ active participation to the study (ePRO), to minimize the burden of in-clinic visits and to explore the potential of this hybrid approach to better monitor the clinical evaluation of the disease, and treatment effectiveness.

Due to a low rate of enrollment, 11 out of 30 planned patients were included in this study, with 10 patients contributing to passive data collection and thus being included in the analysis. The small size of the population hinders the conclusions that can be drawn from this study, particularly concerning passive data, which are characterized by the high variance in the measured variables. The use of the hybrid (digital and clinical) approach in this study was expected to facilitate patient participation and engagement; however, many patients were unexpectedly skeptical about the use of the wearable device and did not feel confident with the digital technology, leading to early termination of enrollment.

Previous studies have clearly established the efficacy of trazodone OAD in patients with MDD, with significant reduction of MADRS scores and improvements in the quality of sleep among the benefits of treatment (26–28, 35). Despite the low number of patients, data obtained via the three methods employed in this study, i.e., passive data collection, active data collection, and traditional in-clinic data collection, allowed a preliminary digital and clinical phenotyping of patients treated with trazodone OAD monotherapy.

Traditional in-clinic efficacy evaluation (MADRS score), data collected actively (with the exception of the medication intake score), and the sleep score (derived from passively collected data) were coherent and indicated a decrease in the extent of the symptoms of depression following the start of monotherapy with trazodone OAD, confirming its effectiveness in the improvement of depressive symptoms using less conventional methods. Moreover, actively collected parameters such as depression, sleep, warning signs, and anxiety scores, showed an improved clinical outcome starting from Week 2 of the study.

The strength and direction of the correlations between data collected passively and MADRS score were affected by intra- and interpatient variability during the observation period. Trazodone OAD is a multimodal antidepressant, with recognized activity on depressive symptoms and sleep disturbances associated with MDD. Albeit weak, a relationship between the duration of deep sleep, the duration of light sleep, and MADRS score was observed. A higher sleep score, indicating better sleep quality, was correlated with lower MADRS score, suggesting that better sleep quality may be associated with a better clinical outcome of depression.

During this 8-week study, consistent correlations between active parameters (depression, sleep, warning signs, and anxiety scores, assessed weekly by the patients) and MADRS scores were observed. These correlations suggest that an online questionnaire may be a useful tool for close monitoring of patients, outside of scheduled visits and to allow patients to be engaged, providing direct feedback about therapy. This approach has the additional advantage of being completed by patients at home, without the need for a clinic visit.

The COVID-19 pandemic highlighted the ability of digital technology to replace face-to-face interactions, with schoolchildren, work colleagues, and family members adopting or increasing usage of chats and digital platforms to communicate. The adoption of telemedicine (38), particularly for managing outpatient visits during the pandemic (39) may have significant effects on the increasing use of smartphones, apps, and wearable sensors to monitor health (40). The pandemic is also likely to have affected patients’ willingness to attend busy clinical settings (40, 41), an effect which may continue beyond concerns relating only to COVID-19. With increasing access to smartphones (5, 42), and a willingness by patients to use smartphones to monitor their mental health (6, 7), data collected passively and actively using smartphones and wearable devices are likely to contribute to transforming mental healthcare (43).

The diagnosis and assessment of the severity of symptoms of mental health disorders, such as MDD, are not driven by blood tests, radiologic findings, or electrophysiological measurements (2, 44). Instead, diagnosis and treatment outcome are largely based on clinical evaluations and observations, and patients’ subjective descriptions and recall of data on symptoms, and behavioral and daily life functional changes. (11, 44, 45). This is particularly true for patients with depression, for whom reporting symptoms and feelings experienced over a period of time to the caregiver during a short in-clinic visit may be difficult (44, 45).

The development and increased use of wearable devices (40, 44), as well as adherence and interest in using such devices (43, 45), point to benefits of using digital phenotyping and ePRO alongside traditional clinic visits. Data collected passively will likely have a high degree of accuracy, provided patients are given clear instructions and the device is used correctly. Data collected through regular questionnaires via an online platform will not be subject to the recall ability of the patient over a long period of time. Decentralized clinical trials, which enable activities to be carried out at a time and place convenient to the patient, or without any input from the patient beyond wearing and charging a device, reduce patient burden. This is seen as a step toward overcoming a major barrier to recruitment and retention (40, 46). In the same way, it would be hoped that reducing the burden to the patient of monitoring and reporting symptoms would improve accuracy of symptoms’ severity and ease treatment adherence, eventually improving clinical outcomes.

To increase adoption of digital phenotyping and active data collection (via apps or web-based platforms), privacy concerns of patients must be assuaged. Knowledge about mobile sensing apps has been found to be key for participants’ comfortability and intention to the use the app (43). Additionally, participants must feel in control of what data are being shared, and with whom. Interestingly, participants who have received mental health treatment have been shown to be more comfortable with sharing data collected via remote sensing apps than those who have not received mental health treatment, possibly due to a trusting relationship developed over time with their mental health professionals (43).

Despite the limited number of patients enrolled in this clinical study, we believe that the coherence of the trends observed between the passive and active data collection and the final clinical evaluation, a hybrid approach merits further investigation in larger, controlled clinical trials. Digital tools, when combined with the traditional medical practices and evaluation, could improve patient engagement, and provide early information about the treatment response and tolerability, with the benefit of minimal burden on the patient.

We believe that a major benefit of using passively collected data to track the evolution of depressive symptoms is that the burden on the patient can be minimal, compared to traditional data collection methods. Digital phenotyping, via wearable sensors, represents a promising monitoring method to better understand the effect of treatment on the overall functionality of patients and, although limited, the results from this study suggest that further investigation is warranted.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The study received ethical approval from the Etická komise FN Královské Vinohrady (reference number: LEK/04/00/202l) and the Etická komise Research Site s.r.o. (reference number: 219058). Patients provided their written informed consent to participate in this study.

PL, AB, and ACo contributed to the study conception and design. JČ, SP, and AN contributed to the study management and data collection. AR and ACa reviewed the study results. All authors commented on previous versions of the manuscript. All authors contributed to the article and approved the submitted version.

Angelini Pharma S.p.A sponsored this study and supported the writing of the manuscript.

JČ, SP, and AN received principal investigator fees from Angelini Pharma S.p.A. PL, AR, ACo, and ACa are full-time employees of Angelini Pharma S.p.A. AB was a full-time employee of Angelini Pharma S.p.A. at the time of study conduction. The authors declare that this study was sponsored by Angelini Pharma S.p.A. The funder had the following involvement in the study: study design, collection, analysis, interpretation of data, the writing of this article, and the decision to submit it for publication.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.