One-Week Suicide Risk Prediction Using Real-Time Smartphone Monitoring: Prospective Cohort Study.

BACKGROUND: Suicide is a major global public health issue that is becoming increasingly common despite preventive efforts. Though current methods for predicting suicide risk are not sufficiently accurate, technological advances provide invaluable tools with which we may evolve toward a personalized, predictive approach. OBJECTIVE: We aim to predict the short-term (1-week) risk of suicide by identifying changes in behavioral patterns characterized through real-time smartphone monitoring in a cohort of patients with suicidal ideation. METHODS: We recruited 225 patients between February 2018 and March 2020 with a history of suicidal thoughts and behavior as part of the multicenter SmartCrisis study. Throughout 6 months of follow-up, we collected information on the risk of suicide or mental health crises. All participants underwent voluntary passive monitoring using data generated by their own smartphones, including distance walked and steps taken, time spent at home, and app usage. The algorithm constructs daily activity profiles for each patient according to these data and detects changes in the distribution of these profiles over time. Such changes are considered critical periods, and their relationship with suicide-risk events was tested. RESULTS: During follow-up, 18 (8%) participants attempted suicide, and 14 (6.2%) presented to the emergency department for psychiatric care. The behavioral changes identified by the algorithm predicted suicide risk in a time frame of 1 week with an area under the curve of 0.78, indicating good accuracy. CONCLUSIONS: We describe an innovative method to identify mental health crises based on passively collected information from patients' smartphones. This technology could be applied to homogeneous groups of patients to identify different types of crises.

Social media and smartphone app use predicts maintenance of physical activity during Covid-19 enforced isolation in psychiatric outpatients.

There is growing concern that the social and physical distancing measures implemented in response to the Covid-19 pandemic may negatively impact health in other areas, via both decreased physical activity and increased social isolation. Here, we investigated whether increased engagement with digital social tools may help mitigate effects of enforced isolation on physical activity and mood, in a naturalistic study of at-risk individuals. Passively sensed smartphone app use and actigraphy data were collected from a group of psychiatric outpatients before and during imposition of strict Covid-19 lockdown measures. Data were analysed using Gaussian graphical models: a form of network analysis which gives insight into the predictive relationships between measures across timepoints. Within-individuals, we found evidence of a positive predictive path between digital social engagement, general smartphone use, and physical activity-selectively under lockdown conditions (N = 127 individual users, M = 6201 daily observations). Further, we observed a positive relationship between social media use and total daily steps across individuals during (but not prior to) lockdown. Although there are important limitations on the validity of drawing causal conclusions from observational data, a plausible explanation for our findings is that, during lockdown, individuals use their smartphones to access social support, which may help guard against negative effects of in-person social deprivation and other pandemic-related stress. Importantly, passive monitoring of smartphone app usage is low burden and non-intrusive. Given appropriate consent, this could help identify people who are failing to engage in usual patterns of digital social interaction, providing a route to early intervention.

Use of Ecological Momentary Assessment Through a Passive Smartphone-Based App (eB2) by Patients With Schizophrenia: Acceptability Study.

BACKGROUND: Ecological momentary assessment (EMA) tools appear to be useful interventions for collecting real-time data on patients' behavior and functioning. However, concerns have been voiced regarding the acceptability of EMA among patients with schizophrenia and the factors influencing EMA acceptability. OBJECTIVE: The aim of this study was to investigate the acceptability of a passive smartphone-based EMA app, evidence-based behavior (eB2), among patients with schizophrenia spectrum disorders and the putative variables underlying their acceptance. METHODS: The participants in this study were from an ongoing randomized controlled trial (RCT) of metacognitive training, consisting of outpatients with schizophrenia spectrum disorders (F20-29 of 10th revision of the International Statistical Classification of Diseases and Related Health Problems), aged 18-64 years, none of whom received any financial compensation. Those who consented to installation of the eB2 app (users) were compared with those who did not (nonusers) in sociodemographic, clinical, premorbid adjustment, neurocognitive, psychopathological, insight, and metacognitive variables. A multivariable binary logistic regression tested the influence of the above (independent) variables on "being user versus nonuser" (acceptability), which was the main outcome measure. RESULTS: Out of the 77 RCT participants, 24 (31%) consented to installing eB2, which remained installed till the end of the study (median follow-up 14.50 weeks) in 14 participants (70%). Users were younger and had a higher education level, better premorbid adjustment, better executive function (according to the Trail Making Test), and higher cognitive insight levels (measured with the Beck Cognitive Insight Scale) than nonusers (univariate analyses) although only age (OR 0.93, 95% CI 0.86-0.99; P=.048) and early adolescence premorbid adjustment (OR 0.75, 95% CI 0.61-0.93; P=.01) survived the multivariable regression model, thus predicting eB2 acceptability. CONCLUSIONS: Acceptability of a passive smartphone-based EMA app among participants with schizophrenia spectrum disorders in this RCT where no participant received financial compensation was, as expected, relatively low, and linked with being young and good premorbid adjustment. Further research should examine how to increase EMA acceptability in patients with schizophrenia spectrum disorders, in particular, older participants and those with poor premorbid adjustment. TRIAL REGISTRATION: ClinicalTrials.gov NCT04104347; https://clinicaltrials.gov/ct2/show/NCT04104347.

Study protocol of a randomised clinical trial testing whether metacognitive training can improve insight and clinical outcomes in schizophrenia.

BACKGROUND: Although insight in schizophrenia spectrum disorders (SSD) has been associated with positive outcomes, the effect size of previous treatments on insight has been relatively small to date. The metacognitive basis of insight suggests that metacognitive training (MCT) may improve insight and clinical outcomes in SSD, although this remains to be established. METHODS: This single-center, assessor-blind, parallel-group, randomised clinical trial (RCT) aims to investigate the efficacy of MCT for improving insight (primary outcome), including clinical and cognitive insight, which will be measured by the Schedule for Assessment of Insight (Expanded version) (SAI-E) and the Beck Cognitive Scale (BCIS), respectively, in (at least) n = 126 outpatients with SSD at three points in time: i) at baseline (T0); ii) after treatment (T1) and iii) at 1-year follow-up (T2). SSD patients receiving MCT and controls attending a non-intervention support group will be compared on insight level changes and several clinical and cognitive secondary outcomes at T1 and T2, whilst adjusting for baseline data. Ecological momentary assessment (EMA) will be piloted to assess functioning in a subsample of participants. DISCUSSION: To the best of our knowledge, this will be the first RCT testing the effect of group MCT on multiple insight dimensions (as primary outcome) in a sample of unselected patients with SSD, including several secondary outcomes of clinical relevance, namely symptom severity, functioning, which will also be evaluated with EMA, hospitalizations and suicidal behaviour. TRIAL REGISTRATION: ClinicalTrials.gov: NCT04104347. Date of registration: 26/09/2019 (Retrospectively registered).

Combining mobile-health (mHealth) and artificial intelligence (AI) methods to avoid suicide attempts: the Smartcrises study protocol.

BACKGROUND: The screening of digital footprint for clinical purposes relies on the capacity of wearable technologies to collect data and extract relevant information's for patient management. Artificial intelligence (AI) techniques allow processing of real-time observational information and continuously learning from data to build understanding. We designed a system able to get clinical sense from digital footprints based on the smartphone's native sensors and advanced machine learning and signal processing techniques in order to identify suicide risk. METHOD/DESIGN: The Smartcrisis study is a cross-national comparative study. The study goal is to determine the relationship between suicide risk and changes in sleep quality and disturbed appetite. Outpatients from the Hospital Fundación Jiménez Díaz Psychiatry Department (Madrid, Spain) and the University Hospital of Nimes (France) will be proposed to participate to the study. Two smartphone applications and a wearable armband will be used to capture the data. In the intervention group, a smartphone application (MEmind) will allow for the ecological momentary assessment (EMA) data capture related with sleep, appetite and suicide ideations. DISCUSSION: Some concerns regarding data security might be raised. Our system complies with the highest level of security regarding patients' data. Several important ethical considerations related to EMA method must also be considered. EMA methods entails a non-negligible time commitment on behalf of the participants. EMA rely on daily, or sometimes more frequent, Smartphone notifications. Furthermore, recording participants' daily experiences in a continuous manner is an integral part of EMA. This approach may be significantly more than asking a participant to complete a retrospective questionnaire but also more accurate in terms of symptoms monitoring. Overall, we believe that Smartcrises could participate to a paradigm shift from the traditional identification of risks factors to personalized prevention strategies tailored to characteristics for each patient. TRIAL REGISTRATION NUMBER: NCT03720730. Retrospectively registered.

Smartphone-based safety plan for suicidal crisis: The SmartCrisis 2.0 pilot study.

Here we present the findings of the pilot phase of the SmartCrisis 2.0 Randomized Clinical Trial. This pilot study aimed to explore the feasibility and acceptability of a safety plan contained in a smartphone app. Our sample consisted patients with a history of recent suicidal behaviour who installed a smartphone-based safety plan. To explore the satisfaction with of the safety plan, two patient satisfaction surveys were conducted: one qualitative and one quantitative. To explore the objective use of the safety plan, we gained access to texts contained in the safety plans completed by the patients. Participation rate was 77%, while 48.9% patients completed both satisfaction surveys at the end of the pilot phase. N = 105 successfully installed the safety plan. In a scale from 1 to 10, users rated the usefulness of the security plan at 7.4, the usability at 8.9, the degree to which they would recommend it to others at 8.6 and the overall satisfaction with the project including evaluations at 9.6. The most widely completed tab was warning signs. Feeling sad or lonely was the warning sign most commonly reported by patients. The second most completed tab was internal coping strategies. Walking or practicing any other exercise was the strategy most commonly resorted to. Our smartphone-based safety plan appears to be a feasible intervention. Data obtained from this pilot study showed high participation rates and high acceptability by patients. This, together with the general satisfaction with the project, supports its implementation in the clinical practice.

Regularizing transformers with deep probabilistic layers.

Language models (LM) have grown non-stop in the last decade, from sequence-to-sequence architectures to attention-based Transformers. However, regularization is not deeply studied in those structures. In this work, we use a Gaussian Mixture Variational Autoencoder (GMVAE) as a regularizer layer. We study its advantages regarding the depth where it is placed and prove its effectiveness in several scenarios. Experimental result demonstrates that the inclusion of deep generative models within Transformer-based architectures such as BERT, RoBERTa, or XLM-R can bring more versatile models, able to generalize better and achieve improved imputation score in tasks such as SST-2 and TREC or even impute missing/noisy words with richer text.

Multidimensional variability in ecological assessments predicts two clusters of suicidal patients.

The variability of suicidal thoughts and other clinical factors during follow-up has emerged as a promising phenotype to identify vulnerable patients through Ecological Momentary Assessment (EMA). In this study, we aimed to (1) identify clusters of clinical variability, and (2) examine the features associated with high variability. We studied a set of 275 adult patients treated for a suicidal crisis in the outpatient and emergency psychiatric departments of five clinical centers across Spain and France. Data included a total of 48,489 answers to 32 EMA questions, as well as baseline and follow-up validated data from clinical assessments. A Gaussian Mixture Model (GMM) was used to cluster the patients according to EMA variability during follow-up along six clinical domains. We then used a random forest algorithm to identify the clinical features that can be used to predict the level of variability. The GMM confirmed that suicidal patients are best clustered in two groups with EMA data: low- and high-variability. The high-variability group showed more instability in all dimensions, particularly in social withdrawal, sleep measures, wish to live, and social support. Both clusters were separated by ten clinical features (AUC = 0.74), including depressive symptoms, cognitive instability, the intensity and frequency of passive suicidal ideation, and the occurrence of clinical events, such as suicide attempts or emergency visits during follow-up. Initiatives to follow up suicidal patients with ecological measures should take into account the existence of a high variability cluster, which could be identified before the follow-up begins.

Medical Data Wrangling With Sequential Variational Autoencoders.

Medical data sets are usually corrupted by noise and missing data. These missing patterns are commonly assumed to be completely random, but in medical scenarios, the reality is that these patterns occur in bursts due to sensors that are off for some time or data collected in a misaligned uneven fashion, among other causes. This paper proposes to model medical data records with heterogeneous data types and bursty missing data using sequential variational autoencoders (VAEs). In particular, we propose a new methodology, the Shi-VAE, which extends the capabilities of VAEs to sequential streams of data with missing observations. We compare our model against state-of-the-art solutions in an intensive care unit database (ICU) and a dataset of passive human monitoring. Furthermore, we find that standard error metrics such as RMSE are not conclusive enough to assess temporal models and include in our analysis the cross-correlation between the ground truth and the imputed signal. We show that Shi-VAE achieves the best performance in terms of using both metrics, with lower computational complexity than the GP-VAE model, which is the state-of-the-art method for medical records.

Six-Month Clinical and Ecological Momentary Assessment Follow-Up of Patients at High Risk of Suicide: A Survival Analysis.

Objective: In this study, we combined ecological momentary assessment (EMA) with traditional clinical follow-up to explore correlates of suicidal relapse in patients with a history of suicidal behavior.Methods: Over 6 months, we followed up with 393 patients who completed baseline and follow-up interviews and were monitored through smartphone-based EMA via the MEmind app. Recruitment was conducted between February 2018 and March 2020. We recorded the occurrence of clinical suicidal events and EMA suicidal events, the latter defined as extreme scores on questions on passive suicide ideation.Results: Fifteen percent of participants had a new clinical suicidal event during follow-up (9.2% suicide attempt [SA]; 5.9% emergency referral for suicidal ideation [SI]). Of the 319 participants who installed the MEmind app, 20.7% presented with EMA suicidal events. EMA suicidal events were statistically significantly associated with clinical suicidal events at 2-month follow-up but not at 6-month follow-up. In the Cox multivariate regression model, 5 factors were independently associated with clinical suicidal events: number of previous SAs, SA in the past year, SA in the past month (risk factors), female gender, and age (protective factors).Conclusions: Our study confirms some of the risk factors classically associated with risk of suicide reattempt, such as history of suicidal behavior, while questioning others, such as female gender. Risk factors associated with EMA events differed from risk factors associated with traditional clinical suicide events, supporting the existence of distinct suicidal phenotypes.

Real-world feasibility and acceptability of real-time suicide risk monitoring via smartphones: A 6-month follow-up cohort.

Active and passive Ecological Momentary Assessment of suicide risk is crucial for suicide prevention. We aimed to assess the feasibility and acceptability of active and passive smartphone-based EMA in real-world conditions in patients at high risk for suicide. We followed 393 patients at high risk for suicide for six months using two mobile health applications: the MEmind (active) and the eB2 (passive). Retention with active EMA was 79.3% after 1 month and 22.6% after 6 months. Retention with passive EMA was 87.8% after 1 month and 46.6% after 6 months. Satisfaction with the MEmind app, uninstalling the eB2 app and diagnosis of eating disorders were independently associated with stopping active EMA. Satisfaction with the eB2 app and uninstalling the MEmind app were independently associated with stopping passive EMA. Smartphone-based active and passive EMA are feasible and may increase accessibility to mental healthcare.

Smartphone-Based Ecological Momentary Assessment for the Measurement of the Performance Status and Health-Related Quality of Life in Cancer Patients Under Systemic Anticancer Therapies: Development and Acceptability of a Mobile App.

Background: We have defined a project to develop a mobile app that continually records smartphone parameters which may help define the Eastern Cooperative Oncology Group performance status (ECOG-PS) and the health-related quality of life (HRQoL), without interaction with patients or professionals. This project is divided into 3 phases. Here we describe phase 1. The objective of this phase was to develop the app and assess its usability concerning patient characteristics, acceptability, and satisfaction. Methods: The app eB2-ECOG was developed and installed in the smartphone of cancer patients who will be followed for six months. Criteria inclusion were: age over 18-year-old; diagnosed with unresectable or metastatic lung cancer, gastrointestinal stromal tumor, sarcoma, or head and neck cancer; under systemic anticancer therapies; and possession of a Smartphone. The app will collect passive and active data from the patients while healthcare professionals will evaluate the ECOG-PS and HRQoL through conventional tools. Acceptability was assessed during the follow-up. Patients answered a satisfaction survey in the app between 3-6 months from their inclusion. Results: The app developed provides a system for continuously collecting, merging, and processing data related to patient's health and physical activity. It provides a transparent capture service based on all the available data of a patient. Currently, 106 patients have been recruited. A total of 36 patients were excluded, most of them (21/36) due to technological reasons. We assessed 69 patients (53 lung cancer, 8 gastrointestinal stromal tumors, 5 sarcomas, and 3 head and neck cancer). Concerning app satisfaction, 70.4% (20/27) of patients found the app intuitive and easy to use, and 51.9% (17/27) of them said that the app helped them to improve and handle their problems better. Overall, 17 out of 27 patients [62.9%] were satisfied with the app, and 14 of them [51.8%] would recommend the app to other patients. Conclusions: We observed that the app's acceptability and satisfaction were good, which is essential for the continuity of the project. In the subsequent phases, we will develop predictive models based on the collected information during this phase. We will validate the method and analyze the sensitivity of the automated results.

Patients at high risk of suicide before and during a COVID-19 lockdown: ecological momentary assessment study.

The coronavirus disease 2019 (COVID-19) outbreak may have affected the mental health of patients at high risk of suicide. In this study we explored the wish to die and other suicide risk factors using smartphone-based ecological momentary assessment (EMA) in patients with a history of suicidal thoughts and behaviour. Contrary to our expectations we found a decrease in the wish to die during lockdown. This is consistent with previous studies showing that suicide rates decrease during periods of social emergency. Smartphone-based EMA can allow us to remotely assess patients and overcome the physical barriers imposed by lockdown.

Disturbed sleep as a clinical marker of wish to die: A smartphone monitoring study over three months of observation.

BACKGROUND: Smartphone monitoring could contribute to the elucidation of the correlates of suicidal thoughts and behaviors (STB). In this study, we employ smartphone monitoring and machine learning techniques to explore the association of wish to die (passive suicidal ideation) with disturbed sleep, altered appetite and negative feelings. METHODS: This is a prospective cohort study carried out among adult psychiatric outpatients with a history of STB. A daily questionnaire was administered through the MEmind smartphone application. Participants were followed-up for a median of 89.8 days, resulting in 9,878 person-days. Data analysis employed a machine learning technique called Indian Buffet Process. RESULTS: 165 patients were recruited, 139 had the MEmind mobile application installed on their smartphone, and 110 answered questions regularly enough to be included in the final analysis. We found that the combination of wish to die and sleep problems was one of the most relevant latent features found across the sample, showing that these variables tend to be present during the same time frame (96 hours). CONCLUSIONS: Disturbed sleep emerges as a potential clinical marker for passive suicidal ideation. Our findings stress the importance of evaluating sleep as part of the screening for suicidal behavior. Compared to previous smartphone monitoring studies on suicidal behavior, this study includes a long follow-up period and a large sample.

Stability of childhood anxiety disorder diagnoses: a follow-up naturalistic study in psychiatric care.

Few studies have examined the stability of major psychiatric disorders in pediatric psychiatric clinical populations. The objective of this study was to examine the long-term stability of anxiety diagnoses starting with pre-school age children through adolescence evaluated at multiple time points. Prospective cohort study was conducted of all children and adolescents receiving psychiatric care at all pediatric psychiatric clinics belonging to two catchment areas in Madrid, Spain, between 1 January, 1992 and 30 April, 2006. Patients were selected from among 24,163 children and adolescents who received psychiatric care. Patients had to have a diagnosis of an ICD-10 anxiety disorder during at least one of the consultations and had to have received psychiatric care for the anxiety disorder. We grouped anxiety disorder diagnoses according to the following categories: phobic disorders, social anxiety disorders, obsessive-compulsive disorder (OCD), stress-related disorders, and "other" anxiety disorders which, among others, included generalized anxiety disorder, and panic disorder. Complementary indices of diagnostic stability were calculated. As much as 1,869 subjects were included and had 27,945 psychiatric/psychological consultations. The stability of all ICD-10 anxiety disorder categories studied was high regardless of the measure of diagnostic stability used. Phobic and social anxiety disorders showed the highest diagnostic stability, whereas OCD and "other" anxiety disorders showed the lowest diagnostic stability. No significant sex differences were observed on the diagnostic stability of the anxiety disorder categories studied. Diagnostic stability measures for phobic, social anxiety, and "other" anxiety disorder diagnoses varied depending on the age at first evaluation. In this clinical pediatric outpatient sample it appears that phobic, social anxiety, and stress-related disorder diagnoses in children and adolescents treated in community outpatient services may have high diagnostic stability.

Nucleotide variation in central nervous system genes among male suicide attempters.

Despite marked morbidity and mortality associated with suicidal behavior, accurate identification of individuals at risk remains elusive. The goal of this study is to identify a model based on single nucleotide polymorphisms (SNPs) that discriminates between suicide attempters and non-attempters using data mining strategies. We examined functional SNPs (n = 840) of 312 brain function and development genes using data mining techniques. Two hundred seventy-seven male psychiatric patients aged 18 years or older were recruited at a University hospital psychiatric emergency room or psychiatric short stay unit. The main outcome measure was history of suicide attempts. Three SNPs of three genes (rs10944288, HTR1E; hCV8953491, GABRP; and rs707216, ACTN2) correctly classified 67% of male suicide attempters and non-attempters (0.50 sensitivity, 0.82 specificity, positive likelihood ratio = 2.80, negative likelihood ratio = 1.64). The OR for the combined three SNPs was 4.60 (95% CI: 1.31-16.10). The model's accuracy suggests that in the future similar methodologies may generate simple genetic tests with diagnostic utility in identification of suicide attempters. This strategy may uncover new pathophysiological pathways regarding the neurobiology of suicidal acts.

Ecological Momentary Assessment for Monitoring Risk of Suicide Behavior.

In recent years the involvement of technology in psychiatric treatment and its usefulness is increasing. The main advantages of its use lie in the possibility of collecting passively data with greater temporal granularity from each patient individually, since these devices are in direct contact almost every minute of the day with them. The variety of data collected by the all the smartphone sensors allows for a better understanding of the patient behavior through what is called the digital phenotype. So the use of a continuous monitoring system for patients at risk of suicide becomes a very useful tool for improving the quality of life of patients and for the early detection of suicide attempts.

Real-Time Ventricular Cancellation in Unipolar Atrial Fibrillation Electrograms.

Unipolar atrial fibrillation (AF) electrograms (EGMs) require far-field ventricle cancellation to recover hidden atrial activations. Current methods cannot achieve real-time cancellation because of the temporal delay they introduce. We propose a new real-time ventricular cancellation (RVC) method based on causal implementation optimized for real-time functioning. The method is similar to the classical average beat subtraction (ABS) method but it computes the ventricular contribution before the ventricular activation finishes. We compare the proposed method to the ABS on synthetic and real EGM databases for the time and frequency domains. All parameters and their optimal values are analyzed and validated. The RVC method provides a good reconstruction of the unipolar EGMs and a better local activation time detection than the classical approach with average F1scores 0.7307 and 0.7125, respectively. The spectral analysis shows that the average power after ventricular cancellation is reduced for frequency bands between 3 and 5.5 Hz, demonstrating that the proposed method removes the ventricular component present in the unipolar EGM signals compared to the ABS method. The phase mapping analysis on the RVC method presented lower error when comparing the annotated EGM cycles with the phase inversion intervals. In terms of performance ABS and RVC behave similarly, but the real-time capability of the latter justifies its preference over the offline implementations. In the clinical environment other online investigations, e.g., rotational activity assessment, dominant frequency or local activation time mapping, might benefit from the real-time potential of the proposed cancellation method.

A Probabilistic Patient Scheduling Model with Time Variable Slots.

One of the current challenges faced by health centers is to reduce the number of patients who do not attend their appointments. The existence of these patients causes the underutilization of the center's services, which reduces their income and extends patient's access time. In order to reduce these negative effects, several appointment scheduling systems have been developed. With the recent availability of electronic health records, patient scheduling systems that incorporate the patient's no-show prediction are being developed. However, the benefits of including a personalized individual variable time slot for each patient in those probabilistic systems have not been yet analyzed. In this article, we propose a scheduling system based on patients' no-show probabilities with variable time slots and a dynamic priority allocation scheme. The system is based on the solution of a mixed-integer programming model that aims at maximizing the expected profits of the clinic, accounting for first and follow-up visits. We validate our findings by performing an extensive simulation study based on real data and specific scheduling requirements provided by a Spanish hospital. The results suggest potential benefits with the implementation of the proposed allocation system with variable slot times. In particular, the proposed model increases the annual cumulated profit in more than 50% while decreasing the waiting list and waiting times by 30% and 50%, respectively, with respect to the actual appointment scheduling system.

Actigraphic recording of motor activity in depressed inpatients: a novel computational approach to prediction of clinical course and hospital discharge.

Depressed patients present with motor activity abnormalities, which can be easily recorded using actigraphy. The extent to which actigraphically recorded motor activity may predict inpatient clinical course and hospital discharge remains unknown. Participants were recruited from the acute psychiatric inpatient ward at Hospital Rey Juan Carlos (Madrid, Spain). They wore miniature wrist wireless inertial sensors (actigraphs) throughout the admission. We modeled activity levels against the normalized length of admission-'Progress Towards Discharge' (PTD)-using a Hierarchical Generalized Linear Regression Model. The estimated date of hospital discharge based on early measures of motor activity and the actual hospital discharge date were compared by a Hierarchical Gaussian Process model. Twenty-three depressed patients (14 females, age: 50.17 ± 12.72 years) were recruited. Activity levels increased during the admission (mean slope of the linear function: 0.12 ± 0.13). For n = 18 inpatients (78.26%) hospitalised for at least 7 days, the mean error of Prediction of Hospital Discharge Date at day 7 was 0.231 ± 22.98 days (95% CI 14.222-14.684). These n = 18 patients were predicted to need, on average, 7 more days in hospital (for a total length of stay of 14 days) (PTD = 0.53). Motor activity increased during the admission in this sample of depressed patients and early patterns of actigraphically recorded activity allowed for accurate prediction of hospital discharge date.