Ecological Momentary Assessments and Passive Sensing in the Prediction of Short-Term Suicidal Ideation in Young Adults

A systematic review on passive sensing for the prediction of suicidal thoughts and behaviors

Passive sensing data from smartphones and wearables may help improve the prediction of suicidal thoughts and behaviors (STB). In this systematic review, we explored the feasibility and predictive validity of passive sensing for STB. On June 24, 2024, we systematically searched Medline, Embase, Web of Science, PubMed, and PsycINFO. Studies were eligible if they investigated the association between STB and passive sensing, or the feasibility of passive sensing in this context. From 2107 unique records, we identified eleven prediction studies, ten feasibility studies, and seven protocols. Studies indicated generally lower model performance for passive compared to active data, with three out of four studies finding no incremental value. PROBAST ratings revealed major shortcomings in methodology and reporting. Studies suggested that passive sensing is feasible in high-risk populations. In conclusion, there is limited evidence on the predictive value of passive sensing for STB. We highlight important quality characteristics for future research.

Predicting suicidal ideation from irregular and incomplete time series of questionnaires in a smartphone-based suicide prevention platform: a pilot study

Over 700,000 people die by suicide annually. Collecting longitudinal fine-grained data about at-risk individuals, as they occur in the real world, can enhance our understanding of the temporal dynamics of suicide risk, leading to better identification of those in need of immediate intervention. Self-assessment questionnaires were collected over time from 89 at-risk individuals using the EMMA smartphone application. An artificial intelligence (AI) model was trained to assess current level of suicidal ideation (SI), an early indicator of the suicide risk, and to predict its progression in the following days. A key challenge was the unevenly spaced and incomplete nature of the time series data. To address this, the AI was built on a missing value imputation algorithm. The AI successfully distinguished high SI levels from low SI levels both on the current day (AUC = 0.804, F1 = 0.625, MCC = 0.459) and three days in advance (AUC = 0.769, F1 = 0.576, MCC = 0.386). Besides past SI levels, the most significant questions were related to psychological pain, well-being, agitation, emotional tension, and protective factors such as contacts with relatives and leisure activities. This represents a promising step towards early AI-based suicide risk prediction using a smartphone application.

Temporal prediction of suicidal ideation in an ecological momentary assessment study with recurrent neural networks

INTRODUCTION

Ecological Momentary Assessment (EMA) holds promise for providing insights into daily life experiences when studying mental health phenomena. However, commonly used mixed-effects linear statistical models do not fully utilize the richness of the ultidimensional time-varying data that EMA yields. Recurrent Neural Networks (RNNs) provide an alternative data analytic method to leverage more information and potentially improve prediction, particularly for non-normally distributed outcomes.

METHODS

As part of a broader research study of suicidal thoughts and behavior in people with borderline personality disorder (BPD), eighty-four participants engaged in EMA data collection over one week, answering questions multiple times each day about suicidal ideation (SI), stressful events, coping strategy use, and affect. RNNs and mixed-effects linear regression models (MEMs) were trained and used to predict SI. Root mean squared error (RMSE), mean absolute percent error (MAPE), and a pseudo-R2 accuracy metric were used to compare SI prediction accuracy between the two modeling methods.

RESULTS

RNNs had superior accuracy metrics (full model: RMSE = 3.41, MAPE = 42 %, pseudo-R2 = 26 %) compared with MEMs (full model: RMSE = 3.84, MAPE = 56 %, pseudo-R2 = 16 %). Importantly, RNNs showed significantly more accurate prediction at higher values of SI. Additionally, RNNs predicted, with significantly higher accuracy, the SI scores of participants with depression diagnoses and of participants with higher depression scores at baseline.

CONCLUSION

In this EMA study with a moderately sized sample, RNNs were better able to learn and predict daily SI compared with mixed-effects models. RNNs should be considered as an option for EMA analysis.

Preoperative Mobile Health Data Improve Predictions of Recovery From Lumbar Spine Surgery

BACKGROUND AND OBJECTIVES

Neurosurgeons and hospitals devote tremendous resources to improving recovery from lumbar spine surgery. Current efforts to predict surgical recovery rely on one-time patient report and health record information. However, longitudinal mobile health (mHealth) assessments integrating symptom dynamics from ecological momentary assessment (EMA) and wearable biometric data may capture important influences on recovery. Our objective was to evaluate whether a preoperative mHealth assessment integrating EMA with Fitbit monitoring improved predictions of spine surgery recovery.

METHODS

Patients age 21-85 years undergoing lumbar surgery for degenerative disease between 2021 and 2023 were recruited. For up to 3 weeks preoperatively, participants completed EMAs up to 5 times daily asking about momentary pain, disability, depression, and catastrophizing. At the same time, they were passively monitored using Fitbit trackers. Study outcomes were good/excellent recovery on the Quality of Recovery-15 (QOR-15) and a clinically important change in Patient-Reported Outcomes Measurement Information System Pain Interference 1 month postoperatively. After feature engineering, several machine learning prediction models were tested. Prediction performance was measured using the c-statistic.

RESULTS

A total of 133 participants were included, with a median (IQR) age of 62 (53, 68) years, and 56% were female. The median (IQR) number of preoperative EMAs completed was 78 (61, 95), and the median (IQR) number of days with usable Fitbit data was 17 (12, 21). 63 patients (48%) achieved a clinically meaningful improvement in Patient-Reported Outcomes Measurement Information System pain interference. Compared with traditional evaluations alone, mHealth evaluations led to a 34% improvement in predictions for pain interference (c = 0.82 vs c = 0.61). 49 patients (40%) had a good or excellent recovery based on the QOR-15. Including preoperative mHealth data led to a 30% improvement in predictions of QOR-15 (c = 0.70 vs c = 0.54).

CONCLUSION

Multimodal mHealth evaluations improve predictions of lumbar surgery outcomes. These methods may be useful for informing patient selection and perioperative recovery strategies.

Examining missingness at the momentary level in clinical research using ecological momentary assessment: Implications for suicide research

The use of intensive time sampling methods, such as ecological momentary assessment (EMA), has increased in clinical, and specifically suicide, research during the past decade. While EMA can capture dynamic intraindividual processes, repeated assessments increase participant burden, potentially resulting in low compliance. This study aimed to shed light on study-level and psychological variables, including suicidal ideation (SI), that may predict momentary prompt (i.e., prompt-to-prompt) completion. We combined data from three EMA studies examining mental health difficulties (N = 103; 10,656 prompts; 7144 completed), using multilevel models and machine learning to determine how well we can predict prompt-to-prompt completion and which variables are most important. The two most important variables in prompt-to-prompt completion were hours since the last prompt and time in study. Psychological variables added little predictive validity; similarly, trait-level SI demonstrated a small effect on prompt-to-prompt completion. Our study showed how study-level characteristics can be used to explain prompt-to-prompt compliance rates in EMA research, highlighting the potential for developing adaptive assessment schedules to improve compliance.

Examining Passively Collected Smartphone-Based Data in the Days Prior to Psychiatric Hospitalization for a Suicidal Crisis: Comparative Case Analysis

BACKGROUND

Digital phenotyping has seen a broad increase in application across clinical research; however, little research has implemented passive assessment approaches for suicide risk detection. There is a significant potential for a novel form of digital phenotyping, termed screenomics, which captures smartphone activity via screenshots.

OBJECTIVE

This paper focuses on a comprehensive case review of 2 participants who reported past 1-month active suicidal ideation, detailing their passive (ie, obtained via screenomics screenshot capture) and active (ie, obtained via ecological momentary assessment [EMA]) risk profiles that culminated in suicidal crises and subsequent psychiatric hospitalizations. Through this analysis, we shed light on the timescale of risk processes as they unfold before hospitalization, as well as introduce the novel application of screenomics within the field of suicide research.

METHODS

To underscore the potential benefits of screenomics in comprehending suicide risk, the analysis concentrates on a specific type of data gleaned from screenshots-text-captured prior to hospitalization, alongside self-reported EMA responses. Following a comprehensive baseline assessment, participants completed an intensive time sampling period. During this period, screenshots were collected every 5 seconds while one's phone was in use for 35 days, and EMA data were collected 6 times a day for 28 days. In our analysis, we focus on the following: suicide-related content (obtained via screenshots and EMA), risk factors theoretically and empirically relevant to suicide risk (obtained via screenshots and EMA), and social content (obtained via screenshots).

RESULTS

Our analysis revealed several key findings. First, there was a notable decrease in EMA compliance during suicidal crises, with both participants completing fewer EMAs in the days prior to hospitalization. This contrasted with an overall increase in phone usage leading up to hospitalization, which was particularly marked by heightened social use. Screenomics also captured prominent precipitating factors in each instance of suicidal crisis that were not well detected via self-report, specifically physical pain and loneliness.

CONCLUSIONS

Our preliminary findings underscore the potential of passively collected data in understanding and predicting suicidal crises. The vast number of screenshots from each participant offers a granular look into their daily digital interactions, shedding light on novel risks not captured via self-report alone. When combined with EMA assessments, screenomics provides a more comprehensive view of an individual's psychological processes in the time leading up to a suicidal crisis.

Machine Learning for Multimodal Mental Health Detection: A Systematic Review of Passive Sensing Approaches

As mental health (MH) disorders become increasingly prevalent, their multifaceted symptoms and comorbidities with other conditions introduce complexity to diagnosis, posing a risk of underdiagnosis. While machine learning (ML) has been explored to mitigate these challenges, we hypothesized that multiple data modalities support more comprehensive detection and that non-intrusive collection approaches better capture natural behaviors. To understand the current trends, we systematically reviewed 184 studies to assess feature extraction, feature fusion, and ML methodologies applied to detect MH disorders from passively sensed multimodal data, including audio and video recordings, social media, smartphones, and wearable devices. Our findings revealed varying correlations of modality-specific features in individualized contexts, potentially influenced by demographics and personalities. We also observed the growing adoption of neural network architectures for model-level fusion and as ML algorithms, which have demonstrated promising efficacy in handling high-dimensional features while modeling within and cross-modality relationships. This work provides future researchers with a clear taxonomy of methodological approaches to multimodal detection of MH disorders to inspire future methodological advancements. The comprehensive analysis also guides and supports future researchers in making informed decisions to select an optimal data source that aligns with specific use cases based on the MH disorder of interest.