array(42) {
  ["request_overridden_res"]=>
  string(1) "3"
  ["project_status"]=>
  string(7) "ongoing"
  ["project_assoc_trials"]=>
  array(1) {
    [0]=>
    object(WP_Post)#4822 (24) {
      ["ID"]=>
      int(17528)
      ["post_author"]=>
      string(4) "1885"
      ["post_date"]=>
      string(19) "2025-06-23 17:11:46"
      ["post_date_gmt"]=>
      string(19) "2025-06-23 21:11:46"
      ["post_content"]=>
      string(0) ""
      ["post_title"]=>
      string(292) "NCT04338321 - A Long-term Comparison of Esketamine Nasal Spray Versus Quetiapine Extended Release, Both in Combination With a Selective Serotonin Reuptake Inhibitor/​Serotonin-Norepinephrine Reuptake Inhibitor, in Participants With Treatment Resistant Major Depressive Disorder (ESCAPE-TRD)"
      ["post_excerpt"]=>
      string(0) ""
      ["post_status"]=>
      string(7) "publish"
      ["comment_status"]=>
      string(6) "closed"
      ["ping_status"]=>
      string(6) "closed"
      ["post_password"]=>
      string(0) ""
      ["post_name"]=>
      string(188) "nct04338321-a-long-term-comparison-of-esketamine-nasal-spray-versus-quetiapine-extended-release-both-in-combination-with-a-selective-serotonin-reuptake-inhibitor-serotonin-norepinephrine-r"
      ["to_ping"]=>
      string(0) ""
      ["pinged"]=>
      string(0) ""
      ["post_modified"]=>
      string(19) "2025-12-16 12:40:07"
      ["post_modified_gmt"]=>
      string(19) "2025-12-16 17:40:07"
      ["post_content_filtered"]=>
      string(0) ""
      ["post_parent"]=>
      int(0)
      ["guid"]=>
      string(60) "https://yoda.yale.edu/?post_type=clinical_trial&p=17528"
      ["menu_order"]=>
      int(0)
      ["post_type"]=>
      string(14) "clinical_trial"
      ["post_mime_type"]=>
      string(0) ""
      ["comment_count"]=>
      string(1) "0"
      ["filter"]=>
      string(3) "raw"
    }
  }
  ["project_title"]=>
  string(117) "Towards tailored treatment: Predicting real outcomes to augmentation interventions for treatment-resistant depression"
  ["project_narrative_summary"]=>
  string(837) "Depression is common, and for many people, standard treatments do not work well. These people often go through several rounds of different medications before finding a treatment that helps. This project aims to make treatment more personalised to individual people, by developing a tool that can help predict which people are most likely to get better when a second mediation is added. The tool will use information that is routinely collected in healthcare, such as information on an individual’s symptoms and background. To do this, we will analyse data from patients receiving different add-on treatments. We will build statistical models that learn patterns from this data and then test whether these models can make predictions in a separate group of patients. This helps ensure the tool can be used in a wider range of patients. "
  ["project_learn_source"]=>
  string(11) "data_holder"
  ["principal_investigator"]=>
  array(7) {
    ["first_name"]=>
    string(8) "Anthony "
    ["last_name"]=>
    string(6) "Cleare"
    ["degree"]=>
    string(24) "BSc, MBBS, PhD, FRCPsych"
    ["primary_affiliation"]=>
    string(21) "King's College London"
    ["email"]=>
    string(24) "anthony.cleare@kcl.ac.uk"
    ["state_or_province"]=>
    string(6) "London"
    ["country"]=>
    string(14) "United Kingdom"
  }
  ["project_key_personnel"]=>
  array(2) {
    [0]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(5) "Megan"
      ["p_pers_l_name"]=>
      string(7) "Boreham"
      ["p_pers_degree"]=>
      string(4) "MScR"
      ["p_pers_pr_affil"]=>
      string(21) "King's College London"
      ["p_pers_scop_id"]=>
      string(0) ""
      ["requires_data_access"]=>
      string(3) "yes"
    }
    [1]=>
    array(6) {
      ["p_pers_f_name"]=>
      string(8) "Rebecca "
      ["p_pers_l_name"]=>
      string(12) "Strawbridge "
      ["p_pers_degree"]=>
      string(4) "PhD "
      ["p_pers_pr_affil"]=>
      string(21) "King's College London"
      ["p_pers_scop_id"]=>
      string(11) "56208851700"
      ["requires_data_access"]=>
      string(3) "yes"
    }
  }
  ["project_ext_grants"]=>
  array(2) {
    ["value"]=>
    string(3) "yes"
    ["label"]=>
    string(65) "External grants or funds are being used to support this research."
  }
  ["project_funding_source"]=>
  string(468) "This research was funded by the National Institute for Health & Care Research (NIHR) Maudsley Biomedical Research Centre (grant number NIHR203318), UK Medical Research Council, and King’s College London. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care. Additional funding from Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), grant number GRK 2773/1 – 454245598. "
  ["project_date_type"]=>
  string(18) "full_crs_supp_docs"
  ["property_scientific_abstract"]=>
  string(1569) "Background: Treatment-resistant depression is common, affecting approximately one third of patients. Despite various treatment strategies being available to patients, including augmentation, there is a lack of data-driven approaches to tailor clinical decision-making to individuals. We are developing clinical prediction models (CPMs) that can be used to determine if an individual patient is likely to respond to augmentation treatment, given their clinical and sociodemographic background. 

Objectives: To externally validate CPMs of response to augmentation in TRD in an independent dataset, considering response as determined by measures of depressive symptom burden and functioning. 

Study design: Application and assessment of performance of previously specified CPMs in an external validation dataset.

Participants: Adults with treatment-resistant depression who took part in the ESCAPE-TRD randomised clinical trial. 

Outcome Measures: The primary outcome measures for treatment response are the clinician-administered Montgomery- Åsberg Depression Rating Scale (MADRS) and the patient-reported Sheehan Disability Scale (SDS). 

Statistical Analyses: Predictive risk of responding to augmentation treatment will be calculated for each individual. The ability of the models to generalise to this external population will be evaluated on a range of performance metrics covering both calibration and discrimination, alongside explained variance. Performance will be assessed across demographic subgroups to ensure minimal bias. "
  ["project_brief_bg"]=>
  string(2864) "Depression remains a common mental health disorder, accounting for 1.47% of all disability adjusted life years globally in 2019 (1). Part of this burden can be attributed to the high rates of treatment resistance within MDD, with treatment-resistant depression (TRD) being commonly defined as a failure to respond adequately to a minimum of two treatments taken at the appropriate dosage and for a sufficient time (2,3). Treatment options for TRD include extending the current treatment trial, switching to an antidepressant from a different class, or augmentation of the current treatment with a secondary agent (4). Whilst augmentation is generally recommended for patients experiencing a partial response to a well-tolerated treatment (5), prescribers must rely on their own clinical experience when determining if an individual patient will be likely to experience a benefit from augmentation, and subsequently, which augmentation agent should be prescribed. 



Clinical prediction models (CPMs) have the potential to aid this clinical decision-making in a data-driven manner, by making predictions on an individual’s likelihood of responding to a given treatment (6). Despite many models showing promise for use in precision psychiatry, progress in the field has been considerably slower than in other areas of medicine. Key contributing issues to the development and deployment of CPMs for mental health include a high risk of bias introduced by subpar analytical decisions during model development, and significant implementation barriers relating to the associated costs, negative perceptions of CPMs, questions over their potential harm to service users, and a lack of external validation of models in independent datasets (7–9). Within TRD, CPMs have largely focused on predicting onset of resistance (10–14); to our knowledge, only one study has looked at predicting response to next-line treatments for TRD (15). 



Our research project aims to develop a clinically relevant, externally validated CPM for response to augmentation treatment in TRD. The models will be developed using data from the LQD study, a large, long-term, pragmatic, open label, randomised control trial that compared the clinical and cost-effectiveness of lithium and quetiapine as augmentation treatments for TRD (16). To ensure the models are clinically meaningful to patients, the outcomes used to operationalise treatment response will be informed by patient and public involvement (PPI) consultation, and the models will be developed using nested-cross validation to prevent overfitting to the training data (17). We then aim to validate the final models in an external dataset evaluating long-term outcomes of augmentation therapy in TRD, to determine if they can generalise to other healthcare settings and TRD patient populations. 

"
  ["project_specific_aims"]=>
  string(594) "The primary aim of this project is to validate our clinical prediction models in an external cohort of participants with TRD being treated with a first-line augmentation strategy. Model performance will be assessed on a range of metrics, covering both calibration and discrimination, across subpopulations to evaluate model fairness. 



The secondary aim is to assess if these models can generalise to predict longer-term outcomes to more novel pharmacological treatment strategies for TRD, such as esketamine, considering the difference in mechanism of action and speed of onset. "
  ["project_study_design"]=>
  array(2) {
    ["value"]=>
    string(5) "other"
    ["label"]=>
    string(5) "Other"
  }
  ["project_study_design_exp"]=>
  string(63) "External validation of prediction models of treatment response "
  ["project_purposes"]=>
  array(2) {
    [0]=>
    array(2) {
      ["value"]=>
      string(76) "confirm_or_validate previously_conducted_research_on_treatment_effectiveness"
      ["label"]=>
      string(76) "Confirm or validate previously conducted research on treatment effectiveness"
    }
    [1]=>
    array(2) {
      ["value"]=>
      string(50) "research_on_clinical_prediction_or_risk_prediction"
      ["label"]=>
      string(50) "Research on clinical prediction or risk prediction"
    }
  }
  ["project_research_methods"]=>
  string(642) "Only data from this trial is requested. Beyond the inclusion/exclusion criteria of the trial, we plan to externally validate the prediction models in the 340 participants randomised to the quetiapine arm only. All of the 340 unique adult participants will be included in the validation cohort, as we seek to validate models built under ITT conditions.



For the secondary aim of assessing if the models generalise to more rapid-acting treatments, we will assess model performance in the 336 patients randomised to the esketamine arm. As above, all of the unique adult participants will be included in this generalisation cohort. "
  ["project_main_outcome_measure"]=>
  string(2032) "As we aim to validate CPMs of longer-term outcomes, primary analysis will focus on treatment response from baseline to week 32. PPI work carried out by the authors in development of the models, and a scoping review conducted by the authors, suggested that patients value improvements in functioning in addition to relief for depressive symptoms when evaluating response to treatment (18). We therefore plan to use both the MADRS and the SDS as the primary outcome measures. To account for the variation in symptom burden and functioning that may occur within the 32 weeks, we will use the cumulative time in response on each measure. 



To ensure consistency with previous analyses of the trial, cumulative time in response will be calculated using the same approach previously applied to this dataset (19). Response status will be determined at each study visit using the predefined thresholds. Participants with the same response status at consecutive visits will be assumed to have remained in that state throughout the interval, whereas where response status differs between visits, the transition will be assumed to have occurred at the midpoint of the interval. Cumulative time in response will then be calculated by summing the time spent in the response state across all intervals. 



The authors are currently undertaking work to identify the minimal clinically important difference (MCID) on the MADRS in TRD patients from the LQD trial, and the same cut-offs will be used to define response thresholds for this validation (20).  As the SDS was not collected in the LQD trial, we will use a threshold of a 4-point change to define functional response, since this was identified as meaningful to patients in both a post-hoc analysis of esketamine and outpatients with anxiety and depression (21,22). As secondary outcomes, we will investigate the prediction of response at 32 weeks, considering those likely to respond on the MADRS, the SDS, and those likely to respond on both outcomes.  

"
  ["project_main_predictor_indep"]=>
  string(905) "In order to maximise clinical utility of the developed models, only clinical and sociodemographic predictors that can be measured prior to treatment initiation will be included for selection. To reduce model bias and overfitting, the exact predictors selected for consideration in the development models will be identified a priori to training, informed by an updated systematic review of predictors of augmentation currently being completed by the authors (23). The same predictors identified by the models during development with the LQD data will be applied in the validation using the ESCAPE data. Predictors under consideration include: baseline age, gender, age at diagnosis, duration of episode, number of episodes, number of previous antidepressant treatment trials, depression severity, comorbid psychiatric symptoms, education level, employment status, BMI, and concomitant baseline medication. "
  ["project_other_variables_interest"]=>
  string(671) "Ethnicity will be retained using the categories collected in the original trial (unknown or not reported, American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, Black or African American, White, and more than one race). Where subgroup numbers permit, fairness of model performance will be explored across ethnicity groups. Owing to the small numbers within individual minority ethnic groups, an exploratory comparison of White and minority ethnic participants, may also be undertaken. All other archived baseline clinical outcome assessment data (i.e., MINI, IDS-C30, SF-36, EQ-5D, QLDS, WPAI:SHP, PHQ-9, CGI) may also be considered. 

"
  ["project_stat_analysis_plan"]=>
  string(3676) "The candidate models will be developed exclusively using data from the LQD dataset. Models will be either penalised elastic net or random forests, with the optimal models being selected considering the trade-off between performance and clinical utility, including availability of included predictors and interpretability of the model. All models will be developed using nested cross-validation (minimum k-fold = 10), and the final hyperparameter values selected during model development will be fixed before application to the validation sample. The ESCAPE dataset will be used solely for external validation and assessment of generalisability, with no model updating or recalibration being undertaken in the primary analysis. 



Ahead of deploying the developed models to the ESCAPE dataset for external validation, any required data transformations and centring will be applied to the data to ensure predictor definitions align with those used in the training data. We will then generate outcome measures for each of the defined model outcomes (i.e., cumulative time in response for the MADRS and SDS, using the specified MCID response thresholds). The mean and standard deviation of these outcome measurements will be reported, stratified by baseline depression severity rating, and compared to the distribution of outcomes in the LQD model development dataset. Descriptive characteristics for the included participants will be reported as mean and standard deviations for continuous variables, and absolute counts and relative percentages for categorical variables, in line with the existing trial publications (24). The distribution of included predictors will additionally be compared to the development data distribution, to assess dataset shift. 



Predictions will then be made for each individual patient in the ESCAPE dataset using the models developed from LQD. The expected outcomes will be compared to the observed outcomes, and performance will be assessed using the same metrics for assessing calibration and discrimination as used in the model development sample. These include R^2 (explained variation as measure of overall fit), calibration slope, calibration-in-the-large, and ranked correlation. In line with the TRIPOD+AI guidance, model performance will be assessed across demographic subgroups (e.g., ethnicity, gender) to assess model fairness (25). Differences in model performance upon application to the ESCAPE dataset will be interpreted in the context of differences in the patient populations and demographics, predictor distributions, outcome assessment measures, and healthcare setting (26). 



The handling of missing data in external validation of prediction models is still an active area of research (27). Rather than losing power by only including complete cases, we will consider alternative approaches for the handling of both missing predictor and outcome data. For predictors, we will assess performance both with and without multiple imputation, depending on the number and presumed clinical availability of predictors selected in the final models. For the outcome data, we will use multiple imputation to impute scores for missing MADRS or SDS for any of the timepoints to week 32. As a sensitivity analysis, we will use multiple imputation in line with the original study SAP, counting those with missing outcome data at a given assessment timepoint as not meeting the MCID threshold for response. Although we aim to validate the models in the ITT population, to predict pragmatic, real-world outcomes to inform clinical practice, we will undertake further per-protocol sensitivity analyses. "
  ["project_software_used"]=>
  array(2) {
    [0]=>
    array(2) {
      ["value"]=>
      string(1) "r"
      ["label"]=>
      string(1) "R"
    }
    [1]=>
    array(2) {
      ["value"]=>
      string(7) "rstudio"
      ["label"]=>
      string(7) "RStudio"
    }
  }
  ["project_timeline"]=>
  string(286) "Assuming a project start date of October 2026, we would allow 6 months for all analyses, with planned completion by March 2027. We would therefore target June 2027 for initial manuscript submission for the validation manuscript. Results would be reported back to YODA by October 2027.  "
  ["project_dissemination_plan"]=>
  string(892) "The target audience for this research is academics and clinicians working within psychiatry. In the event of successful external validation and subject to further prospective validation, clinicians may use this research to help inform clinical decision-making regarding augmentation treatment, whereas researchers may use this research to inform work into modelling other treatment choices in TRD. We expect to produce one manuscript from the validation, which we target leading psychiatric research and health modelling journals for publication (e.g. Psychological Medicine, Journal of Affective Disorders). We will also disseminate the findings via conference abstract submissions (e.g. British Association for Psychopharmacology). This work will be reported in line with the TRIPOD+AI statement on reporting of external validation studies to ensure transparent reporting of findings (25). "
  ["project_bibliography"]=>
  string(6344) "

Vos, T. et al. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet 396, 1204–1222 (2020).
U.S. Food and Drug Administration. Major Depressive Disorder: Developing Drugs for Treatment Guidance for Industry. https://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm (2018).
European Medicines Agency. Clinical Investigation of Medicinal Products in the Treatment of Depression – Scientific Guideline. www.ema.europa.eu (2013).
Cleare, A. et al. Evidence-based guidelines for treating depressive disorders with antidepressants: A revision of the 2008 British Association for Psychopharmacology guidelines. Journal of Psychopharmacology 29, 459–525 (2015).
Gaynes, B. N. et al. Treating depression after initial treatment failure: Directly comparing switch and augmenting strategies in STAR*D. J. Clin. Psychopharmacol. 32, 114–119 (2012).
Fusar-Poli, P., Hijazi, Z., Stahl, D. & Steyerberg, E. W. The Science of Prognosis in Psychiatry: A Review. JAMA Psychiatry 75, 1280–1288 (2018).
Meehan, A. J. et al. Clinical prediction models in psychiatry: a systematic review of two decades of progress and challenges. Mol. Psychiatry 27, 2700–2708 (2022).
Salazar De Pablo, G. et al. Implementing Precision Psychiatry: A Systematic Review of Individualized Prediction Models for Clinical Practice. Schizophr. Bull. 47, 284–297 (2021).
Chekroud, A. M. et al. The promise of machine learning in predicting treatment outcomes in psychiatry. Science (1979). 383, 164–167 (2024).
Chekroud, A. M. et al. Cross-trial prediction of treatment outcome in depression: A machine learning approach. Lancet Psychiatry 3, 243–250 (2016).
Mehltretter, J. et al. Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data. Computational Psychiatry 4, 61–75 (2020).
Perlis, R. H. A clinical risk stratification tool for predicting treatment resistance in major depressive disorder. Biol. Psychiatry 74, 7–14 (2013).
Nie, Z., Vairavan, S., Narayan, V. A., Ye, J. & Li, Q. S. Predictive modeling of treatment resistant depression using data from STARD and an independent clinical study. PLoS One 13, (2018).
Curtiss, J., Smoller, J. W. & Pedrelli, P. Optimizing precision medicine for second-step depression treatment: A machine learning approach. Psychol. Med. https://doi.org/10.1017/S0033291724000497 (2024) doi:10.1017/S0033291724000497.
Salem, H. et al. Temporal multi-step predictive modeling of remission in major depressive disorder using early stage treatment data; STAR*D based machine learning approach. J. Affect. Disord. 324, 286–293 (2023).
Cleare, A. J. et al. Clinical and cost-effectiveness of lithium versus quetiapine augmentation for treatment-resistant depression: a pragmatic, open-label, parallel-group, randomised controlled superiority trial in the UK. Lancet Psychiatry 12, 276–288 (2025).
Stahl, D. New horizons in prediction modelling using machine learning in older people’s healthcare research. Age Ageing 53, (2024).
Boreham, M. A., Bogdanova, M., Bloomfield, M., Cleare, A. J. & Strawbridge, R. Protocol For a Scoping Review of Patient-Preferred Outcomes in Depressive Disorders. medRxiv 2024.12.20.24319426 (2025) doi:10.1101/2024.12.20.24319426.
Vieta, E. et al. Improvements in functioning and workplace productivity with esketamine nasal spray versus quetiapine extended release in patients with treatment resistant depression: Findings from a 32-week randomised, open-label, rater-blinded phase IIIb study. European Neuropsychopharmacology 93, 29–39 (2025).
Boreham, M., Strawbridge, R., Stahl, D. & Cleare, T. LQD: Determining Thresholds For Clinically Meaningful Change On Depressive Symptomatology And Patient-Reported Outcomes In Patients With Treatment-Resistant Depression. https://doi.org/10.17605/OSF.IO/HFDJT (2025).
Turkoz, I. et al. Clinically meaningful changes on depressive symptom measures and patient-reported outcomes in patients with treatment-resistant depression. Acta Psychiatr. Scand. 143, 253–263 (2021).
Abdin, E. et al. Responsiveness and minimal important differences of common disability measures in people with depression and anxiety disorders. Frontiers in Rehabilitation Sciences 6, (2025).
Boreham, M. A., Boules, C., Strawbridge, R. & Cleare, A. J. Predictors of response to treatment augmentation in patients with treatment resistant depression: an updated systematic review. PROSPERO CRD420251148950 https://www.crd.york.ac.uk/PROSPERO/view/CRD420251148950 (2025).
Reif, A. et al. Esketamine Nasal Spray versus Quetiapine for Treatment-Resistant Depression. New England Journal of Medicine 389, 1298–1309 (2023).
Collins, G. S. et al. TRIPOD+AI statement: Updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ https://doi.org/10.1136/bmj-2023-078378 (2024) doi:10.1136/bmj-2023-078378.
Subbaswamy, A. & Saria, S. From development to deployment: dataset shift, causality, and shift-stable models in health AI. Biostatistics 21, 345–352 (2020).
Riley, R. D. et al. Evaluation of clinical prediction models (part 2): how to undertake an external validation study. BMJ 384, (2024).

"
  ["project_suppl_material"]=>
  bool(false)
  ["project_coi"]=>
  array(3) {
    [0]=>
    array(1) {
      ["file_coi"]=>
      array(21) {
        ["ID"]=>
        int(19740)
        ["id"]=>
        int(19740)
        ["title"]=>
        string(11) "COI FORM AC"
        ["filename"]=>
        string(15) "COI-FORM-AC.pdf"
        ["filesize"]=>
        int(37603)
        ["url"]=>
        string(64) "https://yoda.yale.edu/wp-content/uploads/2026/07/COI-FORM-AC.pdf"
        ["link"]=>
        string(57) "https://yoda.yale.edu/data-request/2026-0656/coi-form-ac/"
        ["alt"]=>
        string(0) ""
        ["author"]=>
        string(4) "1885"
        ["description"]=>
        string(0) ""
        ["caption"]=>
        string(0) ""
        ["name"]=>
        string(11) "coi-form-ac"
        ["status"]=>
        string(7) "inherit"
        ["uploaded_to"]=>
        int(19702)
        ["date"]=>
        string(19) "2026-08-03 21:56:23"
        ["modified"]=>
        string(19) "2026-08-03 21:56:23"
        ["menu_order"]=>
        int(0)
        ["mime_type"]=>
        string(15) "application/pdf"
        ["type"]=>
        string(11) "application"
        ["subtype"]=>
        string(3) "pdf"
        ["icon"]=>
        string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png"
      }
    }
    [1]=>
    array(1) {
      ["file_coi"]=>
      array(21) {
        ["ID"]=>
        int(19741)
        ["id"]=>
        int(19741)
        ["title"]=>
        string(11) "COI FORM MB"
        ["filename"]=>
        string(15) "COI-FORM-MB.pdf"
        ["filesize"]=>
        int(35984)
        ["url"]=>
        string(64) "https://yoda.yale.edu/wp-content/uploads/2026/07/COI-FORM-MB.pdf"
        ["link"]=>
        string(57) "https://yoda.yale.edu/data-request/2026-0656/coi-form-mb/"
        ["alt"]=>
        string(0) ""
        ["author"]=>
        string(4) "1885"
        ["description"]=>
        string(0) ""
        ["caption"]=>
        string(0) ""
        ["name"]=>
        string(11) "coi-form-mb"
        ["status"]=>
        string(7) "inherit"
        ["uploaded_to"]=>
        int(19702)
        ["date"]=>
        string(19) "2026-08-03 21:56:26"
        ["modified"]=>
        string(19) "2026-08-03 21:56:26"
        ["menu_order"]=>
        int(0)
        ["mime_type"]=>
        string(15) "application/pdf"
        ["type"]=>
        string(11) "application"
        ["subtype"]=>
        string(3) "pdf"
        ["icon"]=>
        string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png"
      }
    }
    [2]=>
    array(1) {
      ["file_coi"]=>
      array(21) {
        ["ID"]=>
        int(19733)
        ["id"]=>
        int(19733)
        ["title"]=>
        string(24) "YODA-Project-COI-Form-RS"
        ["filename"]=>
        string(32) "COI-Form-Rebecca-Strawbridge.pdf"
        ["filesize"]=>
        int(37454)
        ["url"]=>
        string(81) "https://yoda.yale.edu/wp-content/uploads/2026/07/COI-Form-Rebecca-Strawbridge.pdf"
        ["link"]=>
        string(74) "https://yoda.yale.edu/data-request/2026-0656/coi-form-rebecca-strawbridge/"
        ["alt"]=>
        string(0) ""
        ["author"]=>
        string(4) "1885"
        ["description"]=>
        string(0) ""
        ["caption"]=>
        string(0) ""
        ["name"]=>
        string(28) "coi-form-rebecca-strawbridge"
        ["status"]=>
        string(7) "inherit"
        ["uploaded_to"]=>
        int(19702)
        ["date"]=>
        string(19) "2026-08-03 19:27:43"
        ["modified"]=>
        string(19) "2026-08-03 19:28:15"
        ["menu_order"]=>
        int(0)
        ["mime_type"]=>
        string(15) "application/pdf"
        ["type"]=>
        string(11) "application"
        ["subtype"]=>
        string(3) "pdf"
        ["icon"]=>
        string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png"
      }
    }
  }
  ["data_use_agreement_training"]=>
  bool(true)
  ["human_research_protection_training"]=>
  bool(true)
  ["certification"]=>
  bool(true)
  ["search_order"]=>
  string(1) "0"
  ["project_send_email_updates"]=>
  bool(false)
  ["project_publ_available"]=>
  bool(true)
  ["project_year_access"]=>
  string(4) "2026"
  ["project_rep_publ"]=>
  bool(false)
  ["project_assoc_data"]=>
  array(0) {
  }
  ["project_due_dil_assessment"]=>
  array(21) {
    ["ID"]=>
    int(20372)
    ["id"]=>
    int(20372)
    ["title"]=>
    string(47) "YODA Project Due Diligence Assessment 2026-0656"
    ["filename"]=>
    string(51) "YODA-Project-Due-Diligence-Assessment-2026-0656.pdf"
    ["filesize"]=>
    int(107902)
    ["url"]=>
    string(100) "https://yoda.yale.edu/wp-content/uploads/2026/07/YODA-Project-Due-Diligence-Assessment-2026-0656.pdf"
    ["link"]=>
    string(93) "https://yoda.yale.edu/data-request/2026-0656/yoda-project-due-diligence-assessment-2026-0656/"
    ["alt"]=>
    string(0) ""
    ["author"]=>
    string(4) "1885"
    ["description"]=>
    string(0) ""
    ["caption"]=>
    string(0) ""
    ["name"]=>
    string(47) "yoda-project-due-diligence-assessment-2026-0656"
    ["status"]=>
    string(7) "inherit"
    ["uploaded_to"]=>
    int(19702)
    ["date"]=>
    string(19) "2026-11-03 16:54:11"
    ["modified"]=>
    string(19) "2026-11-03 16:54:11"
    ["menu_order"]=>
    int(0)
    ["mime_type"]=>
    string(15) "application/pdf"
    ["type"]=>
    string(11) "application"
    ["subtype"]=>
    string(3) "pdf"
    ["icon"]=>
    string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png"
  }
  ["project_title_link"]=>
  array(21) {
    ["ID"]=>
    int(20373)
    ["id"]=>
    int(20373)
    ["title"]=>
    string(46) "YODA Project Protocol - 2026-0656 - 2026-08-05"
    ["filename"]=>
    string(46) "YODA-Project-Protocol-2026-0656-2026-08-05.pdf"
    ["filesize"]=>
    int(184478)
    ["url"]=>
    string(95) "https://yoda.yale.edu/wp-content/uploads/2026/07/YODA-Project-Protocol-2026-0656-2026-08-05.pdf"
    ["link"]=>
    string(88) "https://yoda.yale.edu/data-request/2026-0656/yoda-project-protocol-2026-0656-2026-08-05/"
    ["alt"]=>
    string(0) ""
    ["author"]=>
    string(4) "1885"
    ["description"]=>
    string(0) ""
    ["caption"]=>
    string(0) ""
    ["name"]=>
    string(42) "yoda-project-protocol-2026-0656-2026-08-05"
    ["status"]=>
    string(7) "inherit"
    ["uploaded_to"]=>
    int(19702)
    ["date"]=>
    string(19) "2026-11-03 16:54:31"
    ["modified"]=>
    string(19) "2026-11-03 16:54:31"
    ["menu_order"]=>
    int(0)
    ["mime_type"]=>
    string(15) "application/pdf"
    ["type"]=>
    string(11) "application"
    ["subtype"]=>
    string(3) "pdf"
    ["icon"]=>
    string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png"
  }
  ["project_review_link"]=>
  array(21) {
    ["ID"]=>
    int(20374)
    ["id"]=>
    int(20374)
    ["title"]=>
    string(36) "YODA Project Review - 2026-0656_site"
    ["filename"]=>
    string(38) "YODA-Project-Review-2026-0656_site.pdf"
    ["filesize"]=>
    int(1321216)
    ["url"]=>
    string(87) "https://yoda.yale.edu/wp-content/uploads/2026/07/YODA-Project-Review-2026-0656_site.pdf"
    ["link"]=>
    string(80) "https://yoda.yale.edu/data-request/2026-0656/yoda-project-review-2026-0656_site/"
    ["alt"]=>
    string(0) ""
    ["author"]=>
    string(4) "1885"
    ["description"]=>
    string(0) ""
    ["caption"]=>
    string(0) ""
    ["name"]=>
    string(34) "yoda-project-review-2026-0656_site"
    ["status"]=>
    string(7) "inherit"
    ["uploaded_to"]=>
    int(19702)
    ["date"]=>
    string(19) "2026-11-03 16:54:49"
    ["modified"]=>
    string(19) "2026-11-03 16:54:49"
    ["menu_order"]=>
    int(0)
    ["mime_type"]=>
    string(15) "application/pdf"
    ["type"]=>
    string(11) "application"
    ["subtype"]=>
    string(3) "pdf"
    ["icon"]=>
    string(62) "https://yoda.yale.edu/wp/wp-includes/images/media/document.png"
  }
  ["project_highlight_button"]=>
  string(0) ""
  ["request_data_partner"]=>
  string(0) ""
}
data partner
array(1) {
  [0]=>
  string(0) ""
}


pi country
array(0) {
}


pi affil
array(0) {
}


products
array(0) {
}


num of trials
array(1) {
  [0]=>
  string(1) "0"
}


res
array(1) {
  [0]=>
  string(1) "3"
}