Recent Progress in Biosensors for Depression Monitoring-Advancing Personalized Treatment

Abstract

Depression is currently a major contributor to unnatural deaths and the healthcare burden globally, and a patient's battle with depression is often a long one. Because the causes, symptoms, and effects of medications are complex and highly individualized, early identification and personalized treatment of depression are key to improving treatment outcomes. The development of wearable electronics, machine learning, and other technologies in recent years has provided more possibilities for the realization of this goal. Conducting regular monitoring through biosensing technology allows for a more comprehensive and objective analysis than previous self-evaluations. This includes identifying depressive episodes, distinguishing somatization symptoms, analyzing etiology, and evaluating the effectiveness of treatment programs. This review summarizes recent research on biosensing technologies for depression. Special attention is given to technologies that can be portable or wearable, with the potential to enable patient use outside of the hospital, for long periods.

Keywords: biosensors; depression; personalized treatment.

Figure 1

(a) Depression results from an interaction between multiple risk and protective factors that is unique for each person. Regardless of the distal origins of the causal pathways, they converge on brain development and function and are expressed in multiple brain regions that interact to mediate various depressive features (shown in blue, red, and green). These brain patterns are highly variable, likely reflecting etiological differences, variations in the degree of illness severity and persistence, and the heterogeneous expression of mood, motor, cognitive, and vegetative symptoms among individuals [7]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2021 by the authors, published by Elsevier Ltd. (b) The global prevalence of major depressive disorder (MDD) before and after adjustment for (i.e., during) the COVID-19 pandemic, 2020, by age and sex [8]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2021 by the authors, published by Elsevier Ltd. (c) Global burden (disability-adjusted life-years) of MDD and anxiety disorders by age and sex [8]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2021 by the authors, published by Elsevier Ltd.

Figure 2

(a) Depressive symptoms in diverse global populations [7]. Data from Haroz and colleagues, from 170 study populations and 76 nationalities or ethnicities [19]. * ICD or DSM major depressive episode symptoms. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2021 by the authors, published by Elsevier Ltd. (b) The number of papers on various machine learning classification methods in MDD studies, 2000–2017 [20]. (c) Box plots of 66 articles on MDD recognition accuracy based on five methods [20]. Reproduced with permission, Copyright 2018 John Wiley & Sons Ltd.

Figure 3

Transform devices that would otherwise be difficult to use at home into wearable sensors, which in turn enable convenient monitoring. (a) Flexible substrate turns field effect tube sensors into wearable sensors for cortisol sensing [32]. Reproduced with permission, Copyright 2024 ROYAL SOCIETY OF CHEMISTRY. (b) Microneedling allows for the detection of markers in subcutaneous capillaries, enabling wearable sensors for blood sample sensing [36]. Reproduced with permission, Copyright 2024 American Chemical Society.

Figure 4

MIP method for sensing without consumables. It can be used to develop wearable or portable sensors. (a) Wearable technology to detect cortisol in sweat using MIP technology [37]. Reproduced with permission, Copyright 2024 Wiley-VCH (b), and a miniaturized sensor for thyrotropic protamine [38]. Reproduced with permission, Copyright 2019 Elsevier.

Figure 5

Colorimetry makes it possible to achieve sensing using smartphones and even the human eye. (a) Cortisol sensing using a smartphone camera. (b) Real images of the strip test under a UV lamp at 365 nm before and after cortisol exposure [47]. Reproduced with permission, Copyright 2024 WILEY—V C H VERLAG GMBH & CO. KGAA (c) Colorimetric sensing of melatonin was achieved using blue-emissive carbon dots (BCDs) [48]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2024 by the authors, published by Elsevier B.V. (d) The morphological changes of AuTNPs resulted in vivid color variations of the nanoprism dispersion, accompanied by a blue shift of the in-plane LSPR peak, enabling visual and photometric sensing [49]. The blue and red dotted lines are the original extinction spectra obtained from intact and etched AuTNPs, respectively. Reproduced with permission, Copyright 2019 American Chemical Society.

Figure 6

SPR sensor. (a) The SPR effect on the gold surface produces Goos–Hanchen uniqueness in the presence of markers, which in turn enables sensing of TNF [113]. The curves in the figure are superimposed sensorgram of titrated amount of TNF-α (1 fM to 1 μM) over a time course of 10 minutes. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2023 by the authors, published by Elsevier B.V. (b) TNF with aptamer is attracted to the surface of gold nanolayer for SPR sensing [114]. Reproduced with permission, Copyright 2024 Analyst.

Figure 7

Inflammation-related sensing has important implications for the monitoring of depression. (a) The effect of intestinal inflammation and depression or anxiety [115]. Reproduced with permission, Copyright 2022 Springer Nature Limited. (b) Inflammation-related signaling pathways in the human brain [81]. Reproduced with permission, Copyright 2022 Springer Nature Limited. (c) Real-time monitoring of CRP with the help of cell phone NFC technology [121]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2024 by the authors, published by American Chemical Society.

Figure 8

Applications of carbon-based metamaterials. (a) Laser-induced graphene has a large specific surface area and the prepared electrodes also have good flexibility and can be used for dopamine sensing [96]. Reproduced with permission, Copyright 2024 American Chemical Society (b) and single-walled carbon nanotubes have a large specific surface area and stability and are also often used in the development of electrodes for cortisol sensing [178]. Reproduced with permission, Copyright 2024 WILEY—V C H VERLAG GMBH & CO. KGAA (c) Carbon nanotubes assembled into fibers have good flexibility and can be combined with everyday clothing for cortisol sensing [173]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2023 by the authors, published by Elsevier.

Figure 9

Different ideas for liquid crystal sensors. (a) The aptamer is pre-positioned at the LC-water interface and the LC molecules resume alignment after the marker attracts the aptamer [191]. Reproduced with permission, Copyright 2023 Springer-Verlag GmbH, DE part of Springer Nature. (b) LC molecules are originally well aligned and the entry of marker and aptamer disrupts the molecular alignment [188]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2020 by the authors, published by Elsevier Inc.

Figure 10

Depression markers that require further research on sensing technologies. (a) Gene set enrichment analyses on the genes annotated to significant differentially methylated regions in the longitudinal analysis performed between T0 and T12, in patients who underwent EMDR. Top 10 enriched gene sets for (A) GO BP and (B) GO MF. Top 10 enriched pathways for (C) KEGG and (D) Reactome. (E) Top 10 enriched Transcription Factor Target gene set defined by the MSigDB (collection C3: regulatory target gene sets, GTRD subset) [213] Reproduced under the terms of the Creative Commons Attribution License, Copyright 2024 by the authors, published by Informa UK Limited (b) The altered tendency for GM of CAD and anxiety and depression from the four aspects of Firmicutes, Bacteroidetes, Proteobacteria, and Actinobacteria [214]. Upward and downward arrows indicate an increase or decrease in the number of bacteria in the situation of anxiety and depression, respectively. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2021 by the authors, published by Aging and Disease (c) The correlation analysis showed that GDNF protein level negatively correlated with the value of HAMD-17 in PSD patients (correlation coefficient  =  −0.328, p  =  0.047). The slash is the result of the fit. Abbreviations: GDNF, glial cell line-derived neurotrophic factor; HAMD, Hamilton depression rating scale; PSD, post-stroke depression [210]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2017 by the authors, published by Springer Nature.

Figure 11

Heartbeat and motion sensing focuses on depression and stress monitoring. (a) A vibration sensor attached to the sternum captures changes in cardiac mechanical properties (such as changes in blood pressure, heart rate and other parameters) to estimate the psychological stress of the wearer [240]. Red arrows indicate the direction of stretching. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2023 by the authors, published by Elsevier B.V. (b) Monitoring HRV and GSR signals through a wristband sensor to assess the driver‘s psychological stress. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2023 by the authors, published by MDPI AG; (c) A depression prediction system based on multi-modal sensor data and machine learning algorithm in smart phones [242]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2020 by the authors, published by MDPI, Basel, Switzerland.

Figure 12

Various types of wearable bioelectrical sensing. (a) Skintronics integrates GSR electrodes, skin temperature sensors and small batteries for wireless monitoring and management of human stress levels [259]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2020 by the authors, published by Elsevier B.V. (b) Flex-printed pre-gelled sensor arrays designed for sleep electroencephalography (EEG) acquisition [267]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2022 by the authors, published by frontiers. (c) Large-area, high-density electromyography electrode arrays directly drawn on the skin, suitable for various muscle structures [268]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2023 by the authors, published by Oxford University Press on behalf of National Academy of Sciences.

Figure 13

(a) Total number of studies using each source of phone data. Note. Accel. = Accelerometer; Gyro. = Gyroscope; BT = Bluetooth; Mic. = Microphone. (b) Number of studies using each sensor type to infer high-level behavioral features. Accel. = Accelerometer; Gyro. = Gyroscope; BT = Bluetooth; Mic. = Microphone [301]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2024 by the authors, published by Elsevier Ltd.

Figure 13

(a) Total number of studies using each source of phone data. Note. Accel. = Accelerometer; Gyro. = Gyroscope; BT = Bluetooth; Mic. = Microphone. (b) Number of studies using each sensor type to infer high-level behavioral features. Accel. = Accelerometer; Gyro. = Gyroscope; BT = Bluetooth; Mic. = Microphone [301]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2024 by the authors, published by Elsevier Ltd.

Figure 14

Types of multimodal systems. (a) The skin patch with integrated EOG, EEG and EMG electrodes was used to evaluate sleep quality and apnea, and showed similar performance as PSG in control experiment [286]. Reproduced under the terms of the Creative Commons Attribution License, Copyright 2023 by the authors, published by American Association for the Advancement of Science. (b) A fully integrated autonomous wearable ultrasound patch system capable of continuously monitoring cardiac signals such as central blood pressure, heart rate, and cardiac output for 12 h, without being affected by the wearer’s movements [230,316]. Reproduced with permission, Copyright 2023 Springer Nature America, Inc. (c)The wearable flexible patch can simultaneously monitor pulse waveform, GSR, skin temperature, sweat metabolites (glucose, lactic acid and UA) and electrolytes (Na+, K+ and NH4+) in real time, to quantify the level of psychological stress [230]. Reproduced with permission, Copyright 2024 Springer Nature Limited.