Detection Algorithm for Recurrence or Relapse of Depression Thanks to a Smartwach

Globally, it is estimated that 3.8% of the population suffers from depression, amounting to approximately 280 million people. In France, 12.5% of individuals aged 18-85 years had a history of a major depressive episode (MDE) within the past 12 months in 2021. In 2017, the prevalence of MDE over the previous 12 months was 8.2% among the actively employed French population. Women exhibited a prevalence rate twice as high as men (11.4% vs. 5.3%). Moderate and severe MDEs accounted for 95% of all MDEs, regardless of gender. A first depressive episode may be isolated and resolve spontaneously within six to nine months without specific treatment. However, in most cases, pharmacological treatment is necessary. Generally, after eight weeks of well-conducted pharmacological treatment, one-third of patients achieve full remission of symptoms, another third partial remission, and the remaining third do not respond to treatment. Following a first depressive episode, about half of the patients will experience a relapse (a reactivation of the ongoing depressive episode, not fully resolved during the remission period) or a recurrence (a depressive state occurring after the patient was deemed recovered), with most cases happening within the first six months. The relapse rate increases further with each decompensation. With each additional episode, the risk of relapse or recurrence rises to up to 90% in individuals with three or more episodes. Depressive disorder is the main predictor of suicide, with one million lives lost annually. Depressive episodes are associated with an increased risk of suicide, morbidity, and mortality. Chronic depression exacerbates the impact of other diseases such as cancer and cardiovascular diseases, with approximately 60% of individuals experiencing severe and lasting impairment in their functioning. More than half of depressive patients will develop multiple episodes, characterized by emotional suffering and despair. Over the past decades, the efficacy of antidepressant treatments has remained stable. About half of the patients remain depressed after initial treatment with psychological or pharmacological therapies, and treatments reduce only one-third of the disease burden. Detecting depression as early as possible could improve the management of depressive patients, which is currently not feasible. Over the past decade, the development of digital applications and the market introduction of wearable devices have offered new opportunities to continuously evaluate physical and psychological parameters. Detection algorithms do not aim to identify new risk factors; rather, they combine multiple known predictors to estimate individual risk. During a major depressive episode, several physical and physiological parameters (heart activity, sleep quality, daily moods) are disrupted. A more precise understanding of these parameters and their variations would allow better characterization of depressive patients and early detection of relapse/recurrence through monitoring. Dalia is a mobile application designed by caregivers in close collaboration with patients and psychiatrists to continuously collect and monitor physiological data via a smartwatch, with the aim of early detection of signs of relapse or recurrence. Early detection of relapse and recurrence risk in MDEs could enable timely management. Although factors for relapse or recurrence have been widely documented, the threshold for variations in physiological (clinical) parameters in a tool detecting relapse or recurrence has not been established. In this study, a set of clinical variables considered potential factors for detecting relapse or recurrence in MDEs will be measured, and their variations will be assessed. These physiological biomarkers will be collected from patients in remission or recovery from a major depressive episode via a smartwatch. We hypothesize that relapse detected by the Dalia device can be correctly adjusted to relapse diagnosis confirmed by a psychiatrist, with high sensitivity and a low false-negative rate.