CoChat
Sign Up
Cited by 98Suicide and Self-Harm StudiesMental Health Research TopicsPsychiatric care and mental health services

A pilot study using frequent inpatient assessments of suicidal thinking to predict short-term postdischarge suicidal behavior

Open access
Wang, Shirley B., Coppersmith, Daniel D. L., Kleiman, Evan M., Bentley, Kate H., Millner, Alexander J., Fortgang, Rebecca, Mair, Patrick, Dempsey, Walter +2 more·2021·JAMA Network Open
DOI 10.1001/jamanetworkopen.2021.0591PMC PMC7944382
Open this paper in CoChat — free accountOpen PDF

Abstract

Importance: The weeks following discharge from psychiatric hospitalization are the highest-risk period for suicide attempts. Real-time monitoring of suicidal thoughts via smartphone prompts may be more indicative of short-term risk than a single, cross-sectional assessment. Objective: To test whether modeling dynamic changes in real-time suicidal thoughts during psychiatric hospitalization can improve predictions of postdischarge suicide attempts vs using only baseline (ie, admission) data or using the mean level of real-time suicidal thoughts during hospitalization. Design, Setting, and Participants: In this prognostic study, 83 adults recruited from the inpatient psychiatric unit at Massachusetts General Hospital completed ecological momentary assessment surveys of suicidal thinking 4 to 6 times per day during hospitalization as well as brief follow-up surveys assessing suicide attempts at 2 and 4 weeks after discharge. Participants completed at least 3 real-time monitoring surveys. Inclusion criteria included hospitalization for suicidal thoughts and/or behaviors and English fluency. Data were collected from January 2016 to December 2018 and analyzed from January to December 2020. Main Outcomes and Measures: The primary outcome was suicide attempt in the month after discharge. Results: Of 83 participants (mean [SD] age, 38.4 [13.6] years; 43 [51.8%] male participants; 69 [83.1%] White individuals), 9 (10.8%) made a suicide attempt in the month after discharge. Mean cross-validated AUC for elastic net models revealed predictive accuracy was fair for the model using baseline data (area under the curve [AUC], 0.71; first to third quartile, 0.55-0.88), good for the model using the mean level of real-time suicidal thoughts during hospitalization (AUC, 0.81; first to third quartile, 0.67-0.91), and best for the model using dynamic changes in real-time suicidal thoughts during hospitalization (AUC, 0.89; first to third quartile, 0.81-0.97); this pattern of results held for other classification metrics (eg, accuracy, positive predictive value, Brier score) and when using different cross-validation procedures. Features assessing rapid fluctuations in suicidal thinking emerged as the strongest predictors of posthospital suicide attempts. A final set of models incorporating percentage missingness further improved both the mean (mean AUC, 0.93; first to third quartile, 0.90-1.00) and dynamic feature (mean AUC, 0.93; first to third quartile, 0.88-1.00) models. Conclusions and Relevance: In this study, collecting real-time data about suicidal thinking during the course of hospitalization significantly improved short-term prediction of posthospitalization suicide attempts. Models including dynamic changes in suicidal thinking over time yielded the best prediction; features that captured rapid changes in suicidal thoughts were particularly strong predictors. Survey noncompletion also emerged as an important predictor of posthospitalization suicide attempts.

AI snapshot

Written from the paper’s full text.

TL;DR

In 83 adults hospitalized for suicidal thoughts and/or behaviors, ecological momentary assessment of suicidal thinking during hospitalization predicted postdischarge suicide attempts better than baseline or mean-level models. The dynamic-change model had mean cross-validated AUC 0.89, and adding EMA missingness improved mean and dynamic models to mean AUC 0.93.

Key findings

  • Of 83 participants, 9 (10.8%) made a suicide attempt in the month after discharge. Mean cross-validated AUC was 0.71 for the baseline model, 0.81 for the mean real-time suicidal thoughts model, and 0.89 for the dynamic feature model.
  • Features capturing rapid fluctuations in suicidal thinking were the strongest predictors; the most important were probability of acute change in desire to kill oneself, intent to kill oneself, and ability to resist the urge to kill oneself.
  • Adding EMA missingness as a predictor improved both the mean model (mean AUC, 0.93; first to third quartile, 0.90–1.00) and dynamic feature model (mean AUC, 0.93; first to third quartile, 0.88–1.00).
  • The dynamic feature model had mean PPV 0.39, compared with 0.18 for the baseline model and 0.28 for the mean model; mean EMA compliance was 52.2% (SD, 18.7%).

Methods

  • Prognostic study of 83 adults recruited from the inpatient psychiatric unit at Massachusetts General Hospital; inclusion required hospitalization for suicidal thoughts and/or behaviors and English fluency, and participants completed at least 3 EMA surveys. Data were collected January 2016–December 2018 and analyzed January–December 2020.
  • Participants completed EMA surveys 4 to 6 times per day during hospitalization and follow-up surveys at 2 and 4 weeks after discharge; suicide attempts were assessed by follow-up self-report and electronic medical record review.
  • Elastic net models predicted suicide attempt in the month after discharge using baseline SITBI data, mean EMA suicidal thought items, or 24 time-series features per EMA item; performance was evaluated primarily with mean cross-validated AUC using 5-fold cross-validation with 3 repetitions.

Limitations

  • Stated: the sample was relatively small (83 participants) and predominantly White (83.1%); future studies need larger, more diverse samples and longer follow-up, especially given the low base rate of suicide attempts.
  • Stated: only self-reported suicidal thoughts/urges were included; compliance was relatively low (mean, 52.2%) and missingness may not have been at random; only 65 of 83 participants completed follow-up, and the authors lacked claims/admissions data from all other area hospitals.
  • Evident from the text: this was a single-site pilot with only 9 postdischarge suicide attempts and no external validation, so model performance estimates are based on cross-validation within a small sample.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1001/jamanetworkopen.2021.0591

Cites

Statistics

Nothing to recompute.

Claims the paper makes

0 of 24 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    The baseline model using admission data had fair predictive accuracy for postdischarge suicide attempt, with mean cross-validated AUC 0.71 (IQR 0.55-0.88).

    “predictive accuracy was fair for the model using baseline data (area under the curve [AUC], 0.71; first to third quartile, 0.55–0.88)”
    • Also stated in: § Model Performance · plain
      “The baseline model provided fair AUC (mean cross-validated AUC, 0.71; interquartile range [IQR], 0.55 to 0.88)”
    Quote in the paperSubstantial claim
  • Findingown data

    Among 83 adults hospitalized for suicidal thoughts and/or behaviors, 9 (10.8%) made a suicide attempt in the month after discharge.

    “Of 83 participants (mean [SD] age, 38.4 [13.6] years; 43 [51.8%] male participants; 69 [83.1%] White individuals), 9 (10.8%) made a suicide attempt in the month after discharge.”
    • Also stated in: § Demographic and Clinical Characteristics · plain
      “In the month following discharge, 9 participants (10.8%) made a suicide attempt; 6 (66.7%) were self-reported on a follow-up survey (with 65 participants [78.3%] completing a follow-up survey), and 3 (33.3%) were identified through electronic medical record review.”
    Quote in the paperSubstantial claim
  • Findingown data

    The dynamic feature model using dynamic changes in real-time suicidal thoughts during hospitalization had the best predictive accuracy, with mean cross-validated AUC 0.89 (IQR 0.81-0.97).

    “best for the model using dynamic changes in real-time suicidal thoughts during hospitalization (AUC, 0.89; first to third quartile, 0.81–0.97)”
    • Also stated in: § Model Performance · plain
      “The dynamic feature model provided a large improvement over the baseline model and further improved on the mean model, with good AUC (mean, 0.89; IQR, 0.81–0.97])”
    Quote in the paperSubstantial claim
  • Findingown data

    Adding percentage missingness as a predictor improved mean AUC for the mean model to 0.93 (IQR 0.90-1.00).

    “A final set of models incorporating percentage missingness further improved both the mean (mean AUC, 0.93; first to third quartile, 0.90–1.00) and dynamic feature (mean AUC, 0.93; first to third quartile, 0.88–1.00) models.”
    • Also stated in: § Secondary Models · plain
      “Adding EMA missingness as a predictor improved mean AUC for both the mean (mean AUC, 0.93; first to third quartile, 0.90–1.00) and dynamic feature (mean AUC, 0.93; first to third quartile, 0.88–1.00) models (Figure 3).”
    Quote in the paperNot a substantial claim
  • Findingown data

    The dynamic feature model's average cross-validated PPV was 0.39.

    “fair PPV (mean, 0.39; IQR, 0.08–0.50)”
    • Also stated in: § Discussion · plain
      “the average cross-validated PPV for our dynamic feature model was 0.39.”
    Quote in the paperSubstantial claim
  • May combine two claims. 10
  • Classified as probably not a substantial claim. 12

  • May combine two claims.

    Findingown data

    EMA missingness emerged as the most important predictor in the mean model and the second most important predictor in the dynamic feature model.

    “In these models, missingness emerged as the second most important predictor for the dynamic feature model, and the most important predictor for the mean model”
    • Also stated in: § Discussion · plain
      “this missingness variable also emerged as the top predictor in the mean model and the second most important predictor in the dynamic feature model.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The pattern of model performance (baseline < mean < dynamic) held for other classification metrics and when using different cross-validation procedures.

    “this pattern of results held for other classification metrics (eg, accuracy, positive predictive value, Brier score) and when using different cross-validation procedures.”
    • Also stated in: § Secondary Models · plain
      “When using different cross-validation procedures to evaluate model robustness, results largely remained unchanged. For both 10-fold cross-validation and leave-one-out-cross-validation, the baseline model provided poor or fair performance, the mean model provided better performance, and the dynamic feature model performed the best”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    Of 104 participants enrolled, 83 (79.8%) completed at least 3 EMA surveys and were included in the analyses.

    “Of 104 participants enrolled, 83 participants (79.8%) completed at least 3 EMA surveys and were included in current analyses.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    The elastic net algorithm was used because penalization improves prediction accuracy and maintains clinical interpretability compared with less transparent machine learning algorithms.

    “We used the elastic net algorithm, given substantial evidence that penalization improves prediction accuracy,25–27 alongside its well-established accuracy and robustness28,29 and ability to maintain clinical interpretability compared with less transparent machine learning algorithms (eg, random forests, neural networks).”
    Quote in the paperSubstantial claim0 of 5 sources checked
    • cites [25] Wu TT, Chen YF, Hastie T, Sobel E, Lange K. Genome-wide association analysis by lasso penalized logistic regression. Bioinformatics. 2009; 25(6): 714– 721. doi: 10.1093/bioinformatics/btp041 19176549 PMC2732298 source

      Not checked in the free run.

    • cites [26] Chung W, Chen J, Turman C, Efficient cross-trait penalized regression increases prediction accuracy in large cohorts using secondary phenotypes. Nat Commun. 2019; 10(1): 569. doi: 10.1038/s41467-019-08535-0 30718517 PMC6361917 source

      Not checked in the free run.

    • cites [27] Haynos AF, Wang SB, Lipson S, Machine learning enhances prediction of illness course: a longitudinal study in eating disorders. Psychol Med. 2020; 1– 11. doi: 10.1017/S0033291720000227 PMC7483262 32108564 source

      Not checked in the free run.

    • cites [28] Ogutu JO, Schulz-Streeck T, Piepho H-P. Genomic selection using regularized linear regression models: ridge regression, lasso, elastic net and their extensions. BMC Proc. 2012; 6 (2)(suppl 2): S10. doi: 10.1186/1753-6561-6-S2-S10 22640436 PMC3363152 source

      Not checked in the free run.

    • cites [29] Zou H, Hastie T. Regularization and variable selection via the elastic net. J Royal Stat Soc Series B (Stat Methodology). 2005; 67(2): 301– 320. doi: 10.1111/j.1467-9868.2005.00503.x source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Patients admitted with suicidal thoughts or behaviors have a suicide rate nearly 200 times the global suicide rate.

    “This risk is particularly heightened among patients admitted with suicidal thoughts or behaviors, with the rate of suicide for these patients being nearly 200 times the global suicide rate.5”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [5] Chung DT, Ryan CJ, Hadzi-Pavlovic D, Singh SP, Stanton C, Large MM. Suicide rates after discharge from psychiatric facilities: a systematic review and meta-analysis. JAMA Psychiatry. 2017; 74(7): 694– 702. doi: 10.1001/jamapsychiatry.2017.1044 28564699 PMC5710249 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Machine learning methods have yielded high AUCs, but the PPV of most models has been low.

    “Although machine learning methods have yielded high AUCs, the PPV of most models has been low.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [36] Belsher BE, Smolenski DJ, Pruitt LD, Prediction models for suicide attempts and deaths: a systematic review and simulation. JAMA Psychiatry. 2019; 76(6): 642– 651. doi: 10.1001/jamapsychiatry.2019.0174 30865249 source

      Not checked in the free run.

  • May combine two claims.

    Study design and conductown data

    The study had a relatively small sample size (83 participants) that was predominantly White (83.1%).

    “First, we had a relatively small sample size (83 participants) that was predominantly White (83.1%).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Only 65 of the original 83 participants completed a follow-up survey.

    “Finally, only 65 of the original 83 participants completed a follow-up survey”
    • Also stated in: § Demographic and Clinical Characteristics · plain
      “with 65 participants [78.3%] completing a follow-up survey”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Suicide attempts are preceded by large, rapid changes in suicidal thinking.

    “prior retrospective work suggests that suicide attempts are preceded by large, rapid changes in suicidal thinking.11,12”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [11] Millner AJ, Lee MD, Nock MK. Describing and measuring the pathway to suicide attempts: a preliminary study. Suicide Life Threat Behav. 2017; 47(3): 353– 369. doi: 10.1111/sltb.12284 27477787 source

      Not checked in the free run.

    • cites [12] Bryan CJ, Rozek DC, Butner J, Rudd MD. Patterns of change in suicide ideation signal the recurrence of suicide attempts among high-risk psychiatric outpatients. Behav Res Ther. 2019; 120: 103392. doi: 10.1016/j.brat.2019.04.001 31104763 PMC7155814 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    A 2017 meta-analysis of 50 years of suicide research found only 1% of predictive analyses used time windows of less than 30 days, usually with just 1 assessment in that window.

    “A 2017 meta-analysis of the past 50 years of suicide research6 revealed only 1% of predictive analyses used time windows of less than 30 days, usually with just 1 assessment in that window.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [6] Franklin JC, Ribeiro JD, Fox KR, Risk factors for suicidal thoughts and behaviors: a meta-analysis of 50 years of research. Psychol Bull. 2017; 143(2): 187– 232. doi: 10.1037/bul0000084 27841450 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductown data

    One participant who made a posthospitalization suicide attempt did not complete the baseline survey, so 82 participants were included in the baseline model.

    “Of note, 1 participant who made a posthospitalization suicide attempt did not complete the baseline survey. Thus, 82 participants were included in this model.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductcitation

    The study used 5-fold cross-validation with 3 repetitions to select optimal lambda and alpha parameters and to obtain a cross-validation estimate of model performance.

    “Following recommendations by Kuhn and Johnson,31 we used 5-fold cross-validation with 3 repetitions to select the optimal λ (shrinkage) and α (mixing) parameters for each elastic net model. This also allowed us to obtain a cross-validation estimate of model performance.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [31] Kuhn M, Johnson K. Applied Predictive Modeling. Springer-Verlag; 2013. doi: 10.1007/978-1-4614-6849-3 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    The study evaluated model robustness using 10-fold cross-validation repeated 3 times and leave-one-out cross-validation.

    “To evaluate model robustness, we also fit our 3 primary models (baseline, mean, dynamic feature model) using 2 different cross-validation procedures, ie, 10-fold repeated 3 times and leave-one-out-cross-validation.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The study only included self-reported measures of suicidal thoughts and urges.

    “Second, we only included self-reported measures of suicidal thoughts and urges”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Prediction models with low PPVs can still have a positive net benefit if the benefits of interventions with true-positive results outweigh costs of intervening with false-positive results.

    “Prediction models with low PPVs can still have a positive net benefit if the benefits of interventions with true-positive results outweigh costs of intervening with false-positive results.37”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [37] Kessler RC. Clinical epidemiological research on suicide-related behaviors—where we are and where we need to go. JAMA Psychiatry. 2019; 76(8): 777– 778. doi: 10.1001/jamapsychiatry.2019.1238 31188420

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductcitation

    Caution is warranted in interpreting the missingness findings because using missingness for predictive modeling may introduce bias and generalize poorly to new data sets.

    “However, caution is warranted in interpreting these findings, given that some evidence suggests using missingness for predictive modeling may introduce bias into models and generalize poorly to new data sets.38”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [38] van Smeden M, Groenwold RHH, Moons KGM. A cautionary note on the use of the missing indicator method for handling missing data in prediction research. J Clin Epidemiol. 2020; 125: 188– 190. doi: 10.1016/j.jclinepi.2020.06.007 32565213

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation

    The study followed the TRIPOD reporting guideline.

    “This study followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guideline.15”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [15] Collins GS, Reitsma JB, Altman DG, Moons KG; TRIPOD Group. Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD): the TRIPOD statement. Circulation. 2015; 131(2): 211– 219. doi: 10.1161/CIRCULATIONAHA.114.014508 25561516 PMC4297220 source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The dynamic feature model provided higher average AUC than the mean model, suggesting metrics of variability may be more important than central tendency.

    “The dynamic feature model provided higher average AUC than the mean model, suggesting metrics of variability may be more important than central tendency.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The study did not have access to claims and/or admissions data for all other hospitals in the area.

    “we did not have access to claims and/or admissions data for all other hospitals in the area.”
    Quote in the paperNot a substantial claim
19 of 19

  • The paper relies on this without support.

    No studies have tested whether real-time measurement of changes in suicidal thinking can prospectively predict imminent suicide attempts.

    “However, no studies have tested whether real-time measurement of changes in suicidal thinking can prospectively predict imminent suicide attempts, which could inform resource allocation for patients most at risk, in a setting where care is already accessible.13”

  • The paper relies on this without support.

    Clinical decision-making regarding hospital discharges is informed by patients’ self-report of suicidal thinking during specific clinical assessments (eg, therapy appointments).

    “Currently, clinical decision-making regarding hospital discharges is informed by patients’ self-report of suicidal thinking during specific clinical assessments (eg, therapy appointments).”
2 of 2

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Pseudo-random sampling schedules were most frequently used (57%, n = 13), followed by fixed sampling (26%, n = 6), and combined fixed and pseudo-random sampling (13%, n = 3).

    “(pseudo)random sampling schedules were most frequently used [57%, n = 13 (42, 47, 51, 58, 61, 65, 69, 71, 79, 83, 85, 86, 88)], followed by fixed sampling [26%, n = 6 (45, 66, 67, 72, 80, 82)], and protocols that combined both fixed and (pseudo)random sampling [13%, n = 3 (47, 56, 60)].”

    From Don't Miss the Moment: A Systematic Review of Ecological Momentary Assessment in Suicide Research (2022) · cites this paper as [79]

  • Studies frequently (57%, n = 13) reported that participants could provide input about their daily schedules, allowing EMA prompt windows to be adjusted for each participant.

    “Studies frequently (57%, n = 13; (47, 51, 58, 60, 61, 69, 71, 79–81, 85, 86)) reported that participants could provide input about their daily schedules (incl. sleep and wake times), allowing EMA prompt windows to be adjusted for each participant”

    From Don't Miss the Moment: A Systematic Review of Ecological Momentary Assessment in Suicide Research (2022) · cites this paper as [79]

  • Suicidal ideation variability was predictive of the occurrence of a suicide attempt at 1-month follow-up post-discharge, based on a pilot study of 83 adults hospitalized for a suicidal crisis.

    “Suicidal ideation variability (here operationalized as the individual's likelihood of experiencing extreme changes in suicidal ideation from one assessment point to the next) was also predictive of the occurrence of a suicide attempt at 1-month follow-up post-discharge, based on a pilot study of 83 adults hospitalized for a suicidal crisis (79).”

    From Don't Miss the Moment: A Systematic Review of Ecological Momentary Assessment in Suicide Research (2022) · cites this paper as [79]

  • The notion that suicidal ideation variability may index suicide risk is tentatively supported by findings of higher variability among patients with more severe ideation, those with a prior suicide attempt history, and by higher EMA variability predicting attempts at 1-month follow-up.

    “This notion is tentatively supported by findings of higher suicidal ideation variability among patients with more severe suicidal ideation (47, 65, 66), as well as those with a prior suicide attempt history (66), and by higher EMA suicidal ideation variability predicting attempts at 1-month follow-up (79).”

    From Don't Miss the Moment: A Systematic Review of Ecological Momentary Assessment in Suicide Research (2022) · cites this paper as [79]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Median survey count is impossible for 83 participants

    Minor

    Each participant's number of completed surveys is an integer. With 83 participants, the median is the 42nd participant's count and therefore cannot be 18.50.

  • Mean model did not outperform baseline on every metric

    Minor

    The discussion says the mean model performed better than baseline across all classification metrics, but its reported accuracy was lower: 0.75 versus 0.77.

2 of 2

The citation lists haven’t been fetched yet

References 42 · Cited by 98

Authors
Wang, S. B., Coppersmith, D. D. L., Kleiman, E. M., Bentley, K. H., Millner, A. J., Fortgang, R., Mair, P., Dempsey, W., Huffman, J. C., & Nock, M. K.
Affiliations
  • Harvard University (US)
  • Rutgers, The State University of New Jersey (US)
  • University of Michigan (US)
  • Massachusetts General Hospital (US)
Corresponding author
Wang, Shirley B.
ORCID
Venue
JAMA Network Open
Publisher
American Medical Association (AMA)
Published
2021-03-09
Volume
4
Issue
3
Pages
e210591
ISSN
2574-3805 (ISSN-L)
DOI
10.1001/jamanetworkopen.2021.0591
PMC
PMC7944382
OpenAlex
W3133986712
Semantic Scholar
5e7fb993e60dc028a1b3ef3dc8d0ef658e4eaa5a
Cited by
98
References
42
Citations by year
2026: 23 · 2025: 10 · 2024: 18 · 2023: 19 · 2022: 19 · 2021: 8 · 2020: 1
FWCI
10.91
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JAMA Network Open (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
Funders
  • National Science Foundation (1745303)
  • U.S. Department of Defense
  • American Psychological Association
  • U.S. Air Force
  • National Institute of Mental Health (K23MH120436)
  • National Institute on Drug Abuse (R01DA039901)
Topics
  • Social Sciences › Psychology › Clinical Psychology › Suicide and Self-Harm Studies
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Mental Health Research Topics
  • Social Sciences › Psychology › Clinical Psychology › Psychiatric care and mental health services
Keywords
suicidal ideation, suicide attempts, psychiatric hospitalization, ecological momentary assessment, real-time monitoring, risk prediction, elastic net, missing data, suicide prevention, post-discharge follow-up, suicide attempt survivors
MeSH terms
Adult, Female, Hospitalization, Humans, Male, Middle Aged, Patient Discharge, Pilot Projects, Prognosis, Self-Assessment, Time Factors, Suicidal Ideation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review