Major Depressive Disorder (MDD) is a mental disorder representing one of the most important diseases affecting human health and longevity [1]. MDD can lead to disability and suicides and there have been over 800,000 cases of suicides globally [2] probably increasing alongside COVID-19 occurrence [3]. The World Health Organization has estimated that by 2030 an explosion of depression-disabling effects in society will affect health and quality of life. Moreover, MDD presents a heterogeneity of clinical features with up to 250 unique symptoms being recognized. Additionally, each of these clinical determinants can also derive from a disease different from MDD [4]. This, combined with the likely concurrent presence of MDD with other comorbidities (e.g., mental disorders or physical impairments), calls for a precise and accurate diagnosis.
The MDD diagnosis is currently symptom-based and the evaluation of its severity mainly relies on psychological interviews and rating scales. These approaches show weak points since diagnosis is mainly made on subjective experiences and self-reports, often affected by memory biases [5].
The knowledge about MDD biological bases is still confined. Research, based on biomarkers, has not revealed sufficient specificity and sensitivity to be considered in clinical routine [6]. Despite the promising results of newly introduced techniques such as Magnetic Resonance Imaging (MRI) markers or electroencephalography (EEG) analysis of front-cortical patterns [7], their use in clinical practice requires cumbersome and expensive equipment as well as skilled technical personnel.
MDD is a disorder that can cause a variety of emotional and physical problems leading to severe clinical implications [8], more common in young and postgraduate students [9]. For this reason, a quick and reliable method to detect the risk of MDD is needed. In this optic, within the last two decades, the relationship between the motor system and emotional processes has been in the spotlight. Attention has been paid to research on the embodiment effects of MDD which highlights the dual aspect concerning the bodily expression of emotion and the way in which emotions are elaborated [10]. It has been shown how motor behaviour is regulated by emotions, and how it represents an indicator of mental illness [11]. Therefore, mental disorders and embodiment could alter brain processes that are related to cognition and therefore the motor system [12], [13]. In particular, in preliminary work, it was demonstrated how MDD patients show a strong association between psychomotor deterioration and reduced walking speed, vertical head movements and a more slumped posture [10].
Consequently, instrumental movement and gait analysis (GA) represents a solution for objective assessment of MDD symptoms on movements. Moreover, it can be translatable to ecological settings and home environments using wearables [14]. In general, GA is an objective method often adopted in the field of neurology, psychology [12] and orthopaedics [15] covering a crucial role in the evaluation of movement in mood disorders too [16]. Several studies assessed posture and gait using different procedures and instruments [17], such as a computerized walkway, force platform and standardized protocols (e.g., Time Up and Go (TUG) test). The early studies on MDD spatiotemporal gait patterns compared healthy and pathological gait, investigating how mood influenced movement change [18], [19]. High-speed cameras and light barriers were adopted by Lemke et al. to investigate both spatial and temporal gait parameters in MDD and control participants [20]. Controls and depressed differed in gait velocity, which is slower in depressed participants, and it is associated with a shorter stride length. These results were confirmed by Michalak et al., that using stereophotogrammetry, found that participants with MDD present reduced arm swing and vertical head movements causing greater lateral swaying movements of the upper body [10]. Gait speed tests present a wide variety of methodologies from study to study, thus making a general description of MDD patients gait patterns very difficult to obtain [21]. In Biggs et al., the association between the risk of falls and depression in older adults has been investigated and a worse TUG performance was found in depression subject [22].
Current research on gait assessment tools is focusing on Inertial Measurement Unit (IMU) sensors [23]. To our knowledge, the use of such technology has not yet been adopted for the investigation of MDD severity. This kind of sensor has been widely exploited for other pathologies such as Parkinson’s disease, cognitive decline [24] and clinical application [25]. IMUs represent an available, inexpensive, and easy-to-use technology which can be embedded in daily-use smart devices and produce immediately translatable results.
For these reasons, in this study, we investigated the signals from IMU-equipped wearable sensors in extracting objective gait parameters, during the TUG test, in both MDD patients and age-matched controls. We focused on gait complexity, viewed in terms of multiscale entropy (MSE). Our hypothesis is that movement complexity can potentially be used as a predictor of MDD occurrence. MSE has been used for analysing brain consciousness, electromyographic signals and EEG signals [26]. The Complexity Index (CI) which is a derived measure from MSE, is a recognized measure of the complexity of physiologic and physical signals of finite length [27]. MSE and CI found applications in evaluating bio-signals and to investigate the difference between healthy and pathological conditions. Lun Hsieh et al. showed how MSE and CI, applied to ground reaction force signals, can differentiate healthy participants from Parkinson’s disease patients [28]. CI has also found a place in the assessement of multiple sclerosis (SM), as demonstrated by Etzemuller et al., who investigated the CI of posturographic signals to compare SM patients with controls [29]. We therefore consider CI as a candidate to be analyzed in retrieving information on MDD assessment too.