Personalised depression forecasting using mobile sensor data and ecological momentary assessment.

Introduction: Digital health interventions are an effective way to treat depression, but it is still largely unclear how patients' individual symptoms evolve dynamically during such treatments. Data-driven forecasts of depressive symptoms would allow to greatly improve the personalisation of treatments. In current forecasting approaches, models are often trained on an entire population, resulting in a general model that works overall, but does not translate well to each individual in clinically heterogeneous, real-world populations. Model fairness across patient subgroups is also frequently overlooked. Personalised models tailored to the individual patient may therefore be promising. Methods: We investigate different personalisation strategies using transfer learning, subgroup models, as well as subject-dependent standardisation on a newly-collected, longitudinal dataset of depression patients undergoing treatment with a digital intervention ( N = 65 patients recruited). Both passive mobile sensor data as well as ecological momentary assessments were available for modelling. We evaluated the models' ability to predict symptoms of depression (Patient Health Questionnaire-2; PHQ-2) at the end of each day, and to forecast symptoms of the next day. Results: In our experiments, we achieve a best mean-absolute-error (MAE) of 0.801 (25% improvement) for predicting PHQ-2 values at the end of the day with subject-dependent standardisation compared to a non-personalised baseline ( MAE = 1.062 ). For one day ahead-forecasting, we can improve the baseline of 1.539 by 12 % to a MAE of 1.349 using a transfer learning approach with shared common layers. In addition, personalisation leads to fairer models at group-level. Discussion: Our results suggest that personalisation using subject-dependent standardisation and transfer learning can improve predictions and forecasts, respectively, of depressive symptoms in participants of a digital depression intervention. We discuss technical and clinical limitations of this approach, avenues for future investigations, and how personalised machine learning architectures may be implemented to improve existing digital interventions for depression.

Journaling Data for Daily PHQ-2 Depression Prediction and Forecasting.

Digital health applications are becoming increasingly important for assessing and monitoring the wellbeing of people suffering from mental health conditions like depression. A common target of said applications is to predict the results of self-assessed Patient-Health-Questionnaires (PHQ), indicating current symptom severity of depressive individuals. Many of the currently available approaches to predict PHQ scores use passive data, e.g., from smartphones. However, there are several other scores and data besides PHQ, e.g., the Behavioral Activation for Depression Scale-Short Form (BADSSF), the Center for Epidemiologic Studies Depression Scale (CESD), or the Personality Dynamics Diary (PDD), all of which can be effortlessly collected on a daily basis. In this work, we explore the potential of using actively-collected data to predict and forecast daily PHQ-2 scores on a newly-collected longitudinal dataset. We obtain a best MAE of 1.417 for daily prediction of PHQ-2 scores, which specifically in the used dataset have a range of 0 to 12, using leave-one-subject-out cross-validation, as well as a best MAE of 1.914 for forecasting PHQ-2 scores using data from up to the last 7 days. This illustrates the additive value that can be obtained by incorporating actively-collected data in a depression monitoring application.

Depression Diagnosis and Forecast based on Mobile Phone Sensor Data.

Previous studies have shown the correlation be-tween sensor data collected from mobile phones and human depression states. Compared to the traditional self-assessment questionnaires, the passive data collected from mobile phones is easier to access and less time-consuming. In particular, passive mobile phone data can be collected on a flexible time interval, thus detecting moment-by-moment psychological changes and helping achieve earlier interventions. Moreover, while previous studies mainly focused on depression diagnosis using mobile phone data, depression forecasting has not received sufficient attention. In this work, we extract four types of passive features from mobile phone data, including phone call, phone usage, user activity, and GPS features. We implement a long short-term memory (LSTM) network in a subject-independent 10-fold cross-validation setup to model both a diagnostic and a forecasting tasks. Experimental results show that the forecasting task achieves comparable results with the diagnostic task, which indicates the possibility of forecasting depression from mobile phone sensor data. Our model achieves an accuracy of 77.0 % for major depression forecasting (binary), an accuracy of 53.7 % for depression severity forecasting (5 classes), and a best RMSE score of 4.094 (PHQ-9, range from 0 to 27).

Modification in the Motor Skills of Seniors in Care Homes Using Serious Games and the Impact of COVID-19: Field Study.

BACKGROUND: The pandemic has highlighted the importance of low-threshold opportunities for exercise and physical activity. At the beginning of 2020, the COVID-19 pandemic led to many restrictions, which affected seniors in care facilities in the form of severe isolation. The isolation led, among other things, to a lack of exercise, which has led to a multitude of negative effects for this target group. Serious games can potentially help by being used anywhere at any time to strengthen skills with few resources. OBJECTIVE: The aim of this study is to evaluate the effectiveness of a serious game to strengthen motor skills (study 1) and the influence of pandemic restrictions (study 2) on seniors in care facilities. METHODS: The data on motor skills (measured by the Tinetti test) originated from an intervention study with repeated measurements that was interrupted by the pandemic conditions. Data were collected 4 times every 3 months with an intervention group (IG, training 3 times for 1 hour per week) and a control group (CG, no intervention). There were 2 substudies. The first considered the first 6 months until the pandemic restrictions, while the second considered the influence of the restrictions on motor skills. RESULTS: The sample size was 70. The IG comprised 31 (44%) participants, with 22 (71%) female and 9 (29%) male seniors with an average age of 85 years. The CG comprised 39 (56%) participants, with 31 (79%) female and 8 (21%) male seniors with an average age of 87 years. In study 1, mixed-design ANOVA showed no significant interaction between measurement times and group membership for the first measurements (F2.136=1.414, P<.25, partial η2=.044), but there was a significant difference between the CG (mean 16.23, SD 1.1) and the IG (mean 19.81, SD 1.2) at the third time of measurement (P=.02). In study 2 the mixed-design ANOVA (used to investigate motor skills before and after the pandemic conditions between the 2 groups) couldn't reveal any significant interaction between measurement times and group membership: F1.67=2.997, P<.09, partial η2=.043. However, there was a significant main effect of the time of measurement: F1.67=5.44, P<.02, partial η²=.075. CONCLUSIONS: During the first 6 months, the IG showed increased motor skills, whereas the motor skills of the CG slightly deteriorated and showed a statistically significant difference after 6 months. The pandemic restrictions leveled the difference and showed a significant negative effect on motor skills over 3 months. As our results show, digital games have the potential to break down access barriers and promote necessary maintenance for important skills. The pandemic has highlighted the importance of low-threshold opportunities for exercise and physical activity. This potentially great benefit for the challenges of tomorrow shows the relevance of the topic and demonstrates the urgent need for action and research. TRIAL REGISTRATION: Deutsches Register klinischer Studien DRKS00016633; https://tinyurl.com/yckmj4px.

The Effectiveness of a Serious Game (MemoreBox) for Cognitive Functioning Among Seniors in Care Facilities: Field Study.

BACKGROUND: Serious games have been found to have enhancing and preventative effects on cognitive abilities in healthy older adults. Yet, there are few results on the effects in older seniors with age-related low cognitive impairments. Their special needs were considered when designing and using innovate technology in the area of prevention, which is especially relevant owing to the continuously aging population. OBJECTIVE: The objective of this study was to evaluate the impact of a serious game on the cognitive abilities of seniors in order to potentially implement innovative resource-oriented technological interventions that can help to meet future challenges. METHODS: In this controlled trial, we tested the serious game MemoreBox, which features modules specifically designed for seniors in nursing homes. Over a period of 1 year, we tested the cognitive abilities of 1000 seniors at 4 time points using the Mini-Mental Status Test. Only half of the participating seniors engaged with the serious game. RESULTS: The study included an intervention group (n=56) and a control group (did not play; n=55). Based on the in-game data collection, a second intervention group (n=38) was identified within the original intervention group, which exactly followed the planned protocol. There were no noteworthy differences between the demographic and main variables of the overall sample. The large reduction in the sample size was due to the effects of the COVID-19 pandemic (drop-out rate: 88.9%). The CI was set at 5%. Mixed analysis of variance (ANOVA) between the cognitive abilities of the intervention and control groups did not show a statistically significant difference between time and group (F2.710,295.379=1.942; P=.13; partial η²=0.018). We noted approximately the same findings for mixed ANOVA between the cognitive abilities of the second intervention and control groups (F3,273=2.574; P=.054; partial η²=0.028). However, we did observe clear tendencies and a statistically significant difference between the 2 groups after 9 months of the intervention (t88.1=-2.394; P=.02). CONCLUSIONS: The results of this study show similarities with the current research situation. Moreover, the data indicate that the intervention can have an effect on the cognitive abilities of seniors, provided that they regularly play the serious game of MemoreBox. The small sample size means that the tendency toward improvement cannot be proven as statistically significant. However, the tendency shown warrants further research. Establishing an effective prevention tool as part of standard care in nursing homes by means of an easy-to-use serious game would be a considerable contribution to the weakened health care system in Germany as it would offer a means of activating senior citizens in partially and fully inpatient care facilities. TRIAL REGISTRATION: German Clinical Trials Register DRKS00016633; https://tinyurl.com/2e4765nj.

Machine learning meets partner matching: Predicting the future relationship quality based on personality traits.

To what extent is it possible to use machine learning to predict the outcome of a relationship, based on the personality of both partners? In the present study, relationship satisfaction, conflicts, and separation (intents) of 192 partners four years after the completion of questionnaires concerning their personality traits was predicted. A 10x10-fold cross-validation was used to ensure that the results of the linear regression models are reproducible. The findings indicate that machine learning techniques can improve the prediction of relationship quality (37% of variance explained), and that the perceived relationship quality of a partner is mostly dependent on his or her own individual personality traits. Additionally, the influences of different sets of variables on predictions are shown: partner and similarity effects did not incrementally predict relationship quality beyond actor effects and general personality traits predicted relationship quality less strongly than relationship-related personality.