Emotion Forecasting: A Transformer-Based Approach (Preprint)

Journal of Medical Internet Research, Journal Year: 2025, Volume and Issue: 27, P. e63962 - e63962

Published: Jan. 16, 2025

Background Monitoring the emotional states of patients with psychiatric problems has always been challenging due to noncontinuous nature clinical assessments, effect health care environment, and inherent subjectivity evaluation instruments. However, mental in disorders exhibit substantial variability over time, making real-time monitoring crucial for preventing risky situations ensuring appropriate treatment. Objective This study aimed leverage new technologies deep learning techniques enable more objective, patients. was achieved by passively variables such as step count, patient location, sleep patterns using mobile devices. We predict self-reports detect sudden variations their valence, identifying that may require intervention. Methods Data this project were collected Evidence-Based Behavior (eB2) app, which records both passive self-reported daily. Passive data refer behavioral information gathered via eB2 app through sensors embedded devices wearables. These obtained from studies conducted collaboration hospitals clinics used eB2. hidden Markov models (HMMs) address missing transformer neural networks time-series forecasting. Finally, classification algorithms applied several variables, including state responses Patient Health Questionnaire-9. Results Through monitoring, we demonstrated ability accurately patients’ anticipate changes time. Specifically, our approach high accuracy (0.93) a receiver operating characteristic (ROC) area under curve (AUC) 0.98 valence classification. For predicting 1 day advance, an ROC AUC 0.87. Furthermore, feasibility forecasting Questionnaire-9, particularly strong performance certain questions. example, question 9, related suicidal ideation, model 0.9 0.77 next day’s response. Moreover, illustrated enhanced stability multivariate when HMM preprocessing combined model, opposed other methods, recurrent or long short-term memory cells. Conclusions The improved methods (eg, network memory), leveraging attention mechanisms capture longer time dependencies gain interpretability. showed potential assess scores questionnaires advance. allows hence better risk detection treatment adjustment.

Language: Английский

---

Advancing digital health equity: Directions for behavioral and social science research

Translational Behavioral Medicine, Journal Year: 2022, Volume and Issue: 13(3), P. 132 - 139

Published: Nov. 1, 2022

The field of digital health is evolving rapidly and encompasses a wide range complex changing technologies used to support individual population health. COVID-19 pandemic has augmented expansion significantly changed how are used. To ensure that these do not create or exacerbate existing disparities, multi-pronged comprehensive research approach needed. In this commentary, we outline five recommendations for behavioral social science researchers critical promoting equity. These include: (i) centering equity in teams theoretical approaches, (ii) focusing on issues literacy engagement, (iii) using methods elevate perspectives needs underserved populations, (iv) ensuring ethical approaches collecting data, (v) developing strategies integrating tools within across systems settings. Taken together, can help advance the justice.The quickly growing changing. Digital have potential increase access health-related information healthcare improve wellbeing, but it important those don’t widen disparities new ones. Behavioral key role play their barriers access, uptake, usage, studying ways voices historically groups, being thoughtful about data collected used, making sure designed be real-world

Language: Английский

---

A Review of AI Cloud and Edge Sensors, Methods, and Applications for the Recognition of Emotional, Affective and Physiological States

Sensors, Journal Year: 2022, Volume and Issue: 22(20), P. 7824 - 7824

Published: Oct. 14, 2022

Affective, emotional, and physiological states (AFFECT) detection recognition by capturing human signals is a fast-growing area, which has been applied across numerous domains. The research aim to review publications on how techniques that use brain biometric sensors can be used for AFFECT recognition, consolidate the findings, provide rationale current methods, compare effectiveness of existing quantify likely they are address issues/challenges in field. In efforts achieve key goals Society 5.0, Industry human-centered design better, affective, progressively becoming an important matter offers tremendous growth knowledge progress these other related fields. this research, sensors, applications was performed, based Plutchik’s wheel emotions. Due immense variety sensing systems, study aimed analysis available define AFFECT, classify them type area their efficiency real implementations. Based statistical multiple criteria 169 nations, our outcomes introduce connection between nation’s success, its number Web Science articles published, frequency citation recognition. principal conclusions present contributes big picture field under explore forthcoming trends.

Language: Английский

---

Challenges in Using mHealth Data From Smartphones and Wearable Devices to Predict Depression Symptom Severity: Retrospective Analysis

Journal of Medical Internet Research, Journal Year: 2023, Volume and Issue: 25, P. e45233 - e45233

Published: Aug. 14, 2023

A number of challenges exist for the analysis mHealth data: maintaining participant engagement over extended time periods and therefore understanding what constitutes an acceptable threshold missing data; distinguishing between cross-sectional longitudinal relationships different features to determine their utility in tracking within-individual variation or screening individuals at high risk; heterogeneity with which depression manifests itself behavioral patterns quantified by passive features. From 479 participants MDD, we extracted 21 capturing mobility, sleep, smartphone use. We investigated impact days available data on feature quality using intraclass correlation coefficient Bland-Altman analysis. then examined nature 8-item Patient Health Questionnaire (PHQ-8) scale (measured every 14 days) individual-mean correlation, repeated measures linear mixed effects model. Furthermore, stratified based difference, features, (depression) low (no depression) PHQ-8 scores Gaussian mixture demonstrated that least 8 (range 2-12) were needed reliable calculation most 14-day window. observed such as sleep onset correlated better cross-sectionally than longitudinally, whereas wakefulness after well longitudinally but worse cross-sectionally. Finally, found could be separated into 3 distinct clusters according difference no depression.

Language: Английский

---

It’s All About Timing: Exploring Different Temporal Resolutions for Analyzing Digital-Phenotyping Data

Advances in Methods and Practices in Psychological Science, Journal Year: 2024, Volume and Issue: 7(1)

Published: Jan. 1, 2024

The use of smartphones and wearable sensors to passively collect data on behavior has great potential for better understanding psychological well-being mental disorders with minimal burden. However, there are important methodological challenges that may hinder the widespread adoption these passive measures. A crucial one is issue timescale: chosen temporal resolution summarizing analyzing affect how results interpreted. Despite its importance, choice rarely justified. In this study, we aim improve current standards digital-phenotyping by addressing time-related decisions faced researchers. For illustrative purposes, from 10 students whose (e.g., GPS, app usage) was recorded 28 days through Behapp application their mobile phones. parallel, participants actively answered questionnaires phones about mood several times a day. We provide walk-through study different timescales doing individualized correlation analyses random-forest prediction models. By so, demonstrate choosing resolutions can lead conclusions. Therefore, propose conducting multiverse analysis investigate consequences resolutions. This will help combat replications crisis caused in part researchers making implicit decisions.

Language: Английский

---

Sensing behavior change in chronic pain: a scoping review of sensor technology for use in daily life

Pain, Journal Year: 2024, Volume and Issue: 165(6), P. 1348 - 1360

Published: Jan. 23, 2024

Technology offers possibilities for quantification of behaviors and physiological changes relevance to chronic pain, using wearable sensors devices suitable data collection in daily life contexts. We conducted a scoping review passive sensor technologies that sample psychological interest including social situations. Sixty articles met our criteria from the 2783 citations retrieved searching. Three-quarters recruited people were with mostly musculoskeletal, remainder acute or episodic pain; those pain had mean age 43 (few studies sampled adolescents children) 60% women. Thirty-seven performed laboratory clinical settings settings. Most used only 1 type technology, 76 types overall. The commonest was accelerometry (mainly contexts), followed by motion capture settings), smaller number collecting autonomic activity, vocal signals, brain activity. Subjective self-report provided "ground truth" mood, other variables, but often at different timescale automatically collected data, many reported weak relationships between technological relevant constructs, instance, fear movement muscle There relatively little discussion practical issues: frequency sampling, missing human reasons, users' experience, particularly when users did not receive any form. conclude some suggestions content process future this field.

Language: Английский

---

Speech Features as Predictors of Momentary Depression Severity in Patients With Depressive Disorder Undergoing Sleep Deprivation Therapy: Ambulatory Assessment Pilot Study

JMIR Mental Health, Journal Year: 2024, Volume and Issue: 11, P. e49222 - e49222

Published: Jan. 18, 2024

Background The use of mobile devices to continuously monitor objectively extracted parameters depressive symptomatology is seen as an important step in the understanding and prevention upcoming episodes. Speech features such pitch variability, speech pauses, rate are promising indicators, but empirical evidence limited, given variability study designs. Objective Previous research studies have found different patterns when comparing single recordings between patients healthy controls, only a few used repeated assessments compare nondepressive episodes within same patient. To our knowledge, no has series measurements with depression (eg, intensive longitudinal data) model dynamic ebb flow subjectively reported concomitant samples. However, data indispensable for detecting ultimately preventing Methods In this study, we captured voice samples momentary affect ratings over course 3 weeks sample (N=30) acute episode receiving stationary care. Patients underwent sleep deprivation therapy, chronotherapeutic intervention that can rapidly improve symptomatology. We hypothesized within-person affective states would be reflected following features: rate. parametrized them using extended Geneva Minimalistic Acoustic Parameter Set (eGeMAPS) from open-source Music Interpretation by Large-Space Extraction (openSMILE; audEERING GmbH) transcript. analyzed along self-reported ratings, multilevel linear regression analysis. average 32 (SD 19.83) per Results Analyses revealed were associated severity, positive affect, valence, energetic arousal; furthermore, pauses negative additionally calmness. Specifically, was negatively improved (ie, lower linked severity well higher arousal). states, whereas positively states. Conclusions Pitch development clinical prediction technologies patient care timely diagnosis monitoring treatment response. Our forward on path developing automated system, facilitating individually tailored treatments increased empowerment.

Language: Английский

---

Identifying depression-related topics in smartphone-collected free-response speech recordings using an automatic speech recognition system and a deep learning topic model

Journal of Affective Disorders, Journal Year: 2024, Volume and Issue: 355, P. 40 - 49

Published: March 27, 2024

Prior research has associated spoken language use with depression, yet studies often involve small or non-clinical samples and face challenges in the manual transcription of speech. This paper aimed to automatically identify depression-related topics speech recordings collected from clinical samples. The data included 3919 English free-response via smartphones 265 participants a depression history. We transcribed automatic recognition (Whisper tool, OpenAI) identified principal transcriptions using deep learning topic model (BERTopic). To risk understand context, we compared participants' severity behavioral (extracted wearable devices) linguistic texts) characteristics across topics. From 29 identified, 6 for depression: 'No Expectations', 'Sleep', 'Mental Therapy', 'Haircut', 'Studying', 'Coursework'. Participants mentioning exhibited higher sleep variability, later onset, fewer daily steps used words, more negative language, leisure-related words their recordings. Our findings were derived depressed cohort specific task, potentially limiting generalizability populations other tasks. Additionally, some had sample sizes, necessitating further validation larger datasets. study demonstrates that can indicate severity. employed data-driven workflow provides practical approach analyzing large-scale real-world settings.

Language: Английский

---

Enhancing the acceptance of smart sensing in psychotherapy patients: findings from a randomized controlled trial

Frontiers in Digital Health, Journal Year: 2024, Volume and Issue: 6

Published: April 18, 2024

Smart sensing has the potential to make psychotherapeutic treatments more effective. It involves passive analysis and collection of data generated by digital devices. However, acceptance smart among psychotherapy patients remains unclear. Based on unified theory use technology (UTAUT), this study investigated (1) toward in a sample (2) effectiveness an facilitating intervention (AFI) (3) determinants acceptance.

Language: Английский

---

A systematic review of engagement reporting in remote measurement studies for health symptom tracking

npj Digital Medicine, Journal Year: 2022, Volume and Issue: 5(1)

Published: June 29, 2022

Remote Measurement Technologies (RMTs) could revolutionise management of chronic health conditions by providing real-time symptom tracking. However, the promise RMTs relies on user engagement, which at present is variably reported in field. This review aimed to synthesise RMT literature identify how and what extent engagement defined, measured, reported, recommendations for standardisation future work. Seven databases (Embase, MEDLINE PsycINFO (via Ovid), PubMed, IEEE Xplore, Web Science, Cochrane Central Register Controlled Trials) were searched July 2020 papers using apps monitoring adults with a condition, prompting users track least three times during study period. Data synthesised critical interpretive synthesis. A total 76 met inclusion criteria. Sixty five percent did not include definition engagement. Thirty included both measurement Four synthetic constructs developed measuring engagement: (i) research protocol, (ii) objective (iii) subjective (iv) interactions between The field currently impeded incoherent measures lack consideration definitions. process implementing reporting design presented, alongside framework options available. Future work should consider as distinct from wider eHealth literature, measure versus engagement.Registration: has been registered PROSPERO [CRD42020192652].

Language: Английский

---