Multivariate Prediction Research Articles

External validation of a multivariable prediction model for positive resection margins in breast-conserving surgery

ObjectivesPositive resection margins after breast-conserving surgery (BCS) most often demands a repeat surgery. To preoperatively identify patients at risk of positive margins, a multivariable model has been developed that predicts positive margins after BCS with a high accuracy. This study aimed to externally validate this prediction model to explore its generalizability and assess if additional preoperatively available variables can further improve its predictive accuracy. The validation cohort included 225 patients with invasive breast cancer who underwent BCS at Aarhus University Hospital, Aarhus, Denmark during 2020–2022. Receiver operating characteristic (ROC) and calibration analysis were used to validate the prediction model. Univariable logistic regression was used to evaluate if additional variables available in the validation cohort were associated with positive margins and backward elimination to explore if these variables could further improve the model´s predictive accuracy.ResultsThe AUC of the model was 0.60 (95% CI: 0.50–0.70) indicating a lower discriminative capacity in the external cohort. We found weak evidence for an association between increased preoperative breast density on mammography and positive resection margins after BCS (p = 0.027), but the AUC of the model did not improve, when mammographic breast density was included as an additional variable in the model.

Pericoronary adipose tissue feature analysis in computed tomography calcium score images in comparison to coronary computed tomography angiography.

We investigated the feasibility and advantages of using non-contrast CT calcium score (CTCS) images to assess pericoronary adipose tissue (PCAT) and its association with major adverse cardiovascular events (MACE). PCAT features from coronary computed tomography angiography (CCTA) have been shown to be associated with cardiovascular risk but are potentially confounded by iodine. If PCAT in CTCS images can be similarly analyzed, it would avoid this issue and enable its inclusion in formal risk assessment from readily available, low-cost CTCS images. To identify coronaries in CTCS images that have subtle visual evidence of vessels, we registered CTCS with paired CCTA images having coronary labels. We developed an "axial-disk" method giving regions for analyzing PCAT features in three main coronary arteries. We analyzed hand-crafted and radiomic features using univariate and multivariate logistic regression prediction of MACE and compared results against those from CCTA. Registration accuracy was sufficient to enable the identification of PCAT regions in CTCS images. Motion or beam hardening artifacts were often prevalent in "high-contrast" CCTA but not CTCS. Mean HU and volume were increased in both CTCS and CCTA for the MACE group. There were significant positive correlations between some CTCS and CCTA features, suggesting that similar characteristics were obtained. Using hand-crafted/radiomics from CTCS and CCTA, AUCs were 0.83/0.79 and 0.83/0.77, respectively, whereas Agatston gave AUC = 0.73. Preliminarily, PCAT features can be assessed from three main coronary arteries in non-contrast CTCS images with performance characteristics that are at the very least comparable to CCTA.

Prediction of pathological complete response after neoadjuvant chemotherapy for HER2-negative breast cancer patients with routine immunohistochemical markers

BackgroundPathological complete response (pCR) is an established surrogate marker for prognosis in patients with breast cancer (BC) after neoadjuvant chemotherapy. Individualized pCR prediction based on clinical information available at biopsy, particularly immunohistochemical (IHC) markers, may help identify patients who could benefit from preoperative chemotherapy.MethodsData from patients with HER2-negative BC who underwent neoadjuvant chemotherapy from 2002 to 2020 (n = 1166) were used to develop multivariable prediction models to estimate the probability of pCR (pCR-prob). The most precise model identified using cross-validation was implemented in an online calculator and a nomogram. Associations among pCR-prob, prognostic IHC3 distant recurrence and disease-free survival were studied using Cox regression and Kaplan–Meier analyses. The model’s utility was further evaluated in independent external validation cohorts.Results273 patients (23.4%) achieved a pCR. The most precise model had across-validated area under the curve (AUC) of 0.84, sensitivity of 0.82, and specificity of 0.71. External validation yielded AUCs between 0.75 (95% CI, 0.70–0.81) and 0.83 (95% CI, 0.78–0.87). The higher the pCR-prob, the greater the prognostic impact of pCR status (presence/absence): hazard ratios decreased from 0.55 (95% central range, 0.07–1.77) at 0% to 0.20 (0.11–0.31) at 50% pCR-prob. Combining pCR-prob and IHC3 score further improved the precision of disease-free survival prognosis.ConclusionsA pCR prediction model for neoadjuvant therapy decision-making was established. Combining pCR and recurrence prediction allows identification of not only patients who benefit most from neoadjuvant chemotherapy, but also patients with a very unfavorable prognosis for whom alternative treatment strategies should be considered.

A multivariable prediction model for invasive pulmonary aspergillosis in immunocompromised patients with acute respiratory failure (IPA-GRRR-OH score).

Invasive pulmonary aspergillosis (IPA) is a life-threatening opportunistic infection in immunocompromised patients. The diagnosis is often made late, with mortality reaching 90% when mechanical ventilation is needed. We sought to develop and validate a risk prediction model for the diagnosis of IPA. We used two independent datasets of immunocompromised patients with acute respiratory failure admitted to 12 intensive care units (ICUs). The derivation dataset include 3262 patients. Factors associated with probable or proven IPA were identified, and a risk prediction model was developed. This model was then validated in a prospective dataset (776 patients). IPA prevalence was 4.5% (146/3262) and 3.3% (26/776), in the derivation and the validation cohorts, respectively. The final model included eight variables constitutive of the IPA-GRRR-OH score: type of immunosuppression, high-dose or long-term corticosteroids, neutropenia, the presence of structural lung disease, time from symptoms onset to ICU admission > 7 days, hemoptysis, focal alveolar pattern on the chest imaging, and viral co-infection. The median score [IQR] was 2 [1-3] in the derivation and 1 [0-3] in the validation cohort. The best cutoff score for IPA diagnosis was 4 (sensitivity 23.1%; specificity 90.5%; negative predictive value 91.4%). Discrimination and calibration were good in both the derivation (AUC 0.72 [0.68-0.76]) and the validation cohort (AUC 0.85 [0.76-0.93]). The IPA-GRRR-OH is a clinical score, easily available at ICU admission, which reliably predicts IPA in immunocompromised patients with acute respiratory failure. Studies to demonstrate benefits from the bedside implementation of this score are warranted.

A multivariable prediction model to stratify risk of 90-day rehospitalization among adults with ulcerative colitis

Abstract Background Individuals with ulcerative colitis (UC) are frequently re-hospitalized for persistent or recurrent severe disease flares. Accurate prediction of the risk of early re-hospitalization at the time of discharge could promote targeted outpatient interventions to reduce this risk. Methods We conducted a retrospective study in adults with UC admitted to The Ottawa Hospital between 2009 and 2016 for an acute UC-related indication. We ascertained candidate demographic, clinical, and health services predictors through medical records and administrative health databases. We derived and bootstrap validated a multivariable logistic regression model of 90-day UC-related re-hospitalization risk. We chose a probability cut point that maximized Youden’s index to differentiate high-risk from low-risk individuals and assessed model performance. Results Among 248 UC-related hospitalizations, there were 27 (10.9%) re-hospitalizations within 90 days of discharge. Our multivariable model identified gastroenterologist consultation within the prior year (adjusted odds ratio [aOR] 0.11, 95% confidence interval [CI], 0.04-0.39), male sex (aOR 3.27, 95% CI, 1.33-8.05), length of stay (OR 0.94, 95% CI, 0.88-1.01), and narcotic prescription at discharge (OR 1.96, 95% CI, 0.73-5.27) as significant predictors of 90-day re-hospitalization. The optimism-corrected c-statistic value was 0.78, and the goodness-of-fit test P-value was .09. The chosen probability cut point produced a sensitivity of 77.8%, specificity of 80.9%, positive predictive value (PPV) of 33.0%, and negative predictive value (NPV) of 96.7% in the derivation cohort. Conclusions A limited set of variables accessible at the point of hospital discharge can reasonably discriminate re-hospitalization risk among individuals with UC. Future studies are required to validate our findings.

Association between obstructive sleep apnea and Tinnitus in the United States: NHANES 2005-2020.

This study aimed to investigate the association between Obstructive Sleep Apnea (OSA) and Tinnitus using NHANES data from 2005 to 2020. This study analyzed data from NHANES (National Health and Nutrition Examination Surveys) conducted between 2005 and 2020, and included 4871 participants aged 16 or older. OSA was assessed using the Multivariate Apnea Prediction Index and the variables from the National Health and Nutrition Examination Survey. Tinnitus was defined as participants who reported being bothered by a ringing, roaring, or buzzing sound in the ears or head lasting 5min or more during the past 12 months. Logistic regression models were employed to examine the association between OSA and Tinnitus. The study cohort had an overall prevalence of Tinnitus of 16.5%, with 53.51% for males and 46.49% for females. After controlling for potential confounders, there was a significant association between Tinnitus and OSA (odds ratio = 1.43, 95% confidence interval = 1.05-1.94, P = 0.03). These findings indicate that OSA is a risk factor for the development of Tinnitus.

DOP093 ECCO Topical Review on predictive models for IBD disease course and treatment response

Abstract Background Inflammatory Bowel Disease (IBD) poses a clinical challenge due to its variable progression and treatment response. Despite the development of predictive models, their clinical application remains limited due to validation and methodological inconsistencies. This ECCO Topical Review examines existing predictive models, assesses their relevance, and discusses the barriers to their clinical implementation. Methods A panel formed by the European Crohn's and Colitis Organisation (ECCO), including gastroenterologists, surgeons, and methodologists, reviewed predictive models on IBD disease course and treatment response. Using the Medline, PubMed, EMBASE, and Scopus databases, they systematically identified relevant literature to evaluate clinical significance, performance, and barriers to clinical integration. Results The panel developed 21 statements on predictive models for IBD disease course and response, with supporting text. Predictive models for IBD disease course For ulcerative colitis (UC), models included clinical factors (e.g., age at diagnosis, male gender), biochemical markers (e.g., C-reactive protein [CRP], anemia), and treatment history. Crohn’s disease (CD) models incorporated disease characteristics like stricturing behavior, perianal disease, and imaging findings. Predictive models for treatment response UC models focused on severe flares and moderate-to-severe cases managed with biologics, with predictors like CRP, disease duration, and endoscopic severity. CD models emphasized predictors like body mass index, smoking status, and CRP levels. The utility of omics data remains inconsistent. Barriers to clinical implementation Models lacked generalizability across diverse populations and face barriers related to clinical integration, healthcare acceptance, regulatory factors, and economics. Methodological approaches Future models should use diverse datasets and adhere to TRIPOD guidelines (1). Validation through metrics like calibration and discrimination, as well as trials, will be essential for real-world effectiveness. Future directions Predictive models should identify treatment responders, aid in safe discontinuation, and support personalized strategies. Models need biomarkers that are cost-effective, reproducible, and adaptable to individual patient profiles. Conclusion Based on this ECCO Topical Review currently being finalized, predictive models for IBD hold promise, but challenges persist in clinical integration. Overcoming methodological issues, performing validation, improving biomarkers, and conducting impact studies are critical. Collaboration among researchers, clinicians, and regulators is pivotal to advance models that improve patient outcomes. References (1)Collins et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. The TRIPOD Group. Circulation 2015;131(2):211–9.

Which variables are associated with recruitment failure? A nationwide review on obstetrical and gynaecological multicentre RCTs (2003–2023)

ObjectiveWe aim to assess which variables are associated with recruitment failure of obstetrical and gynaecological randomised controlled trials (RCTs), leading to an extension of the study period.DesignNationwide study.SettingA cohort of...

Computational pathology identifies a low B-cell content in the tumour microenvironment as a predictor of adverse outcome in patients with classic Hodgkin lymphoma treated with ABVD

AimsThe prognostic impact of B lymphocytes surrounding Hodgkin and Reed Sternberg (HRS) cells in classic Hodgkin lymphoma (cHL) and pathogenic variants in genes associated with apoptosis regulation remains undefined.MethodsWe have...

The External Validation of a Multivariable Prediction Model for Recurrent Pelvic Organ Prolapse After Native Tissue Repair: A Prospective Cohort Study.

Background/Objectives: A prediction model for anatomical cystocele recurrence after native tissue repair was developed and internally validated in 2016. This model estimates a patients' individual risk of recurrence and can be used for counseling. Before implementation in urogynecological clinical practice, external validation is needed. The aim of this study was to assess the external validity of this previously developed prediction model. The secondary aim was to test the performance of this model with a composite and subjective outcome of pelvic organ prolapse (POP) recurrence. Furthermore, the aim was to investigate whether risk factors for POP recurrence were in line with the population in which the original model was developed. Methods: In this prospective multicenter cohort study, 246 patients who underwent anterior colporrhaphy were included. Inclusion criteria were patients scheduled to undergo a primary anterior colporrhaphy (with a POP Quantification (POPQ) stage ≥ 2 cystocele). A combination of a primary anterior colporrhaphy with other POP or incontinence surgery (without the use of vaginal or abdominal mesh material) was permitted. Patients with prolapse or incontinence surgery prior to index surgery could not participate. All patients filled in questionnaires, pelvic floor ultrasound was performed preoperatively, and data from the medical file concerning POPQ stage and obstetric and general history were obtained. Results: Thirty women (12.2%) were lost at follow up. Anatomical cystocele recurrence was present in 107/216 (49.5%), subjective recurrence in 19/208 (9.1%), and 39/219 (17.8%) patients met the criteria for composite outcome. The area under the receiver operating characteristic curves for anatomical, composite, and subjective recurrence were 65.5% (95% CI: 58.7-72.4), 55.8% (95% CI 47.3-64.3%, NS), and 55.1% (95% CI 45.1-65.2%), respectively. In the multivariable analysis, preoperative cystocele stage 3 or 4 and a complete levator defect on ultrasound were independent risk factors for anatomical recurrence. For composite recurrence, younger age and an active employment status were only risk factors in univariable analysis. No significant risk factors for subjective recurrence could be identified. Conclusions: This external validation study showed a moderate performance for a prediction model for anatomical recurrence. The model cannot be used for a composite or subjective outcome prediction because of poor performance. For composite and subjective recurrence, new prediction models need to be developed.

Prediction Models for Postoperative Pneumonia in Elderly Hip Fracture Patients: A Systematic Review and Critical Appraisal.

Although several models have been developed to predict postoperative pneumonia in elderly hip fracture patients, no systematic review of the model quality and clinical applicability has been reported. To systematically review and critically appraise existing models for postoperative pneumonia in elderly hip fracture patients. Systematic review and meta-analysis. 10 databases were systematically searched from inception to April 15, 2024, updated on August 26. Two reviewers independently performed literature selection, information extraction and quality assessment. A narrative synthesis was employed to summarise the characteristics of the models. Meta-analysis was performed using Stata 17.0. 13 studies containing 25 models were included. The prevalence of pneumonia was 9.62% (95% CI: 7.62%-11.62%). Age (53.8%), hypoproteinemia (46.2%), chronic obstructive pulmonary disease (COPD, 30.8%), gender (30.8%), activity of daily living score (ADL, 30.8%) and American Society of Anesthesiologists (ASA, 30.8%) score were the top six predictors. All models reported area under curve (AUC: 0.617-0.996). 9 studies (69.2%) used the Hosmer-Lemeshow (H-L) test, calibration curves, or Brier scores to evaluate the calibration. 5 studies (38.5%) performed internal validation, 4 studies (30.8%) performed external validation. All studies had a high risk of bias due to single sample source, inappropriate data processing, inadequate model evaluation, and negligence of calibration and validation. 10 studies (76.9%) had good applicability. Prediction models for postoperative pneumonia in elderly hip fracture patients are still in the developing stage. The validation and evaluation of existing models are poor. Future studies should focus on robust external validation and updating. Additionally, the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis + artificial intelligence (TRIPOD+AI) statement should be followed. Prediction models are effective in discriminating postoperative pneumonia in elderly hip fracture patients, but further external validation and adjustment are still warranted.

Multivariate grey prediction model with fractional time-lag parameter and its application

Multivariate grey prediction model with fractional time-lag parameter and its application

Multivariable prognostic prediction of efficacy and safety outcomes and response to fingolimod in people with relapsing-remitting multiple sclerosis.

The individual treatment response in people with relapsing-remitting multiple sclerosis (RRMS) remain unpredictable. In order to support medical decisions, we aimed to predict response to fingolimod compared to placebo, by developing and validating prognostic multivariable models. We included two-year follow-up from intention-to-treat populations of two multi-country placebo-controlled randomized controlled trials (RCT) of daily fingolimod 0.5 mg. The data was accessed via ClinicalStudyDataRequest.com (Proposal Number: 11223) The RCTs were in adult RRMS patients with active disease. We used four Cox proportional hazards based penalized (elastic net and grouped lasso) and tree methods (transformation tree and forest) to predict time-to relapse and other relevant efficacy and safety endpoints in data from the RCT FREEDOMS. Treatment arm, 80 baseline variables and their interaction with treatment were considered as candidate predictors in the models. A nested cross-validation scheme ensured independent tuning parameter optimization and internal model performance evaluation. The generalizability of the models with the highest cross-validated time-dependent area under the receiver operating curve (AUC) was further evaluated in terms of discrimination (AUC), calibration (plots, intercept, slope), clinical utility (decision curve analysis), and treatment response plots by external validation in data from the RCT FREEDOMS II. The best performing model predicting relapse risk (331 events) in the development sample (n=843) was an elastic net regression with main terms for four predictors alongside treatment: EDSS score, volume of Gadolinium enhanced T1 lesions, number of relapses in the last 2 years, and number of prior MS treatments. In external validation (n=713), it had an AUC of 0.68 (95% CI 0.63-0.72), but the predictions were overestimating the actual risk (358 events) with a calibration-in-the-large of -0.17 (-0.3 - -0.04) and a slope of 1.06 (0.78-1.35). Almost no heterogeneity (variability 0.001) was detected in the predicted relapse risk change in response to fingolimod. FREEDOMS II participants were predicted to have 0.21 to 0.31 absolute relapse risk reduction with fingolimod compared to placebo. The selected model predicting new or enlarging T2 magnetic resonance imaging (MRI) lesions had an AUC of 0.74 (0.70-0.78), moderate calibration, but no treatment response variability. The final model predicting confirmed disability progression had an AUC of 0.59 (0.54-0.64) and the predicted treatment response heterogeneity was not significant. The overall safety outcome could not be predicted with sufficient discrimination. However, the final model predicting infections or neoplasms had an AUC of 0.69 (0.63-0.74) and non-significant treatment response heterogeneity. For the efficacy outcomes, important predictors were related to (para)clinical disease activity or disability. Unexpected influential predictors included concomitant disorders. Relapse and new or enlarging T2 MRI lesions were moderately predictable in an independent sample with the developed prognostic models. Fingolimod was expected to decrease the risk of these events for all patients, with no predictable heterogeneity. Disability and safety outcomes could not be well-predicted and it is yet unresolved whether the change in their risk as response to fingolimod is heterogeneous or not.

CNN-BiLSTM multivariate load prediction model based on MIC correlation analysis

CNN-BiLSTM multivariate load prediction model based on MIC correlation analysis

Development of a machine learning-based multivariable prediction model for the naturalistic course of generalized anxiety disorder

Development of a machine learning-based multivariable prediction model for the naturalistic course of generalized anxiety disorder

Daily-life walking speed, running duration and bedtime from wrist-worn sensors predict incident dementia: A watch walk - UK biobank study.

To determine if wrist-worn sensor parameters can predict incident dementia in individuals aged 60+years and to compare prediction with other tools. Observational cohort study. Community PARTICIPANTS: The cohort comprised 47,371 participants without dementia, aged 60+years, who participated in the UK Biobank study (mean age=67±4 years; 52% female). Nineteen digital biomarkers were extracted from up-to-7-day wrist-worn sensor accelerometry data at baseline. Univariable and multivariable Cox proportional hazard models examined associations between sensor parameters and prospectively diagnosed dementia. Median follow-up was 7.5 years (interquartile range: 7.0 to 9.0 years), during this time 387 participants (0.8%) were diagnosed with dementia. Among the gait parameters, slower maximal walking speed had the strongest association with incident dementia (32% decrease in hazard for each standard deviation increase) followed by lower daily step counts (30% decrease) and increased step-time variability (17% increase). While adjusting for age and sex, running duration, maximal walking speed and early bedtime were identified as independent and significant predictors of dementia. The multivariable prediction model performed comparably to the ANU-ADRI and UKB-Dementia Risk Score models in the UK Biobank cohort. The study findings indicate that remotely acquired parameters from wrist-worn sensors can predict incident dementia. Since wrist-worn sensors are highly acceptable for long-term use, wrist-worn sensor parameters have the potential to be incorporated into dementia screening programs.

Interpretation of the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis based on regression or machine learning methods (TRIPOD+AI)

Interpretation of the transparent reporting of a multivariable prediction model for individual prognosis or diagnosis based on regression or machine learning methods (TRIPOD+AI)

Prediction of atrial fibrillation after a stroke event: a systematic review with meta-analysis.

Detecting atrial fibrillation (AF) after stroke is a key component of secondary prevention, but indiscriminate prolonged cardiac monitoring is costly and burdensome. Multivariable prediction models could be used to inform patient selection. To determine the performance of available models for predicting AF after a stroke. We searched for studies of multivariable models that were derived, validated and/or augmented for prediction of AF in patients with a stroke, using Medline and Embase from inception through 20/09/2024. Discrimination measures for tools with c-statistic data from ≥3 cohorts were pooled by Bayesian meta-analysis, with heterogeneity assessed through a 95% prediction interval (PI). The risk of bias was assessed using the Prediction Model Risk Of Bias tool (PROBAST). We included 75 studies with 58 prediction models. 66% had a high risk of bias. Fifteen multivariable models were eligible for meta-analysis. Three models showed excellent discrimination: SAFE (c-statistic 0.856, 95% CI 0.796-0.916), SURF (0.815, 95% CI 0.728-0.893), and iPAB (0.888, 95% CI 0.824-0.957). Excluding high-bias studies, only SAFE showed excellent discrimination (0.856, 95% CI 0.800-0.915). No model showed excellent discrimination when limited to external validation or studies with ≥100 AF events. No clinical impact studies were found. Three of the fifty-eight identified multivariable prediction models for AF after stroke demonstrated excellent statistical performance on meta-analysis. However, prospective validation is required to understand the effectiveness of these models in clinical practice before they can be recommended for inclusion in clinical guidelines.

CXCL5 as a biomarker for early diagnosis and prognosis of sepsis: A comprehensive clinical evaluation.

Sepsis, a critical condition caused by a dysregulated host response to infection, has high morbidity and mortality rates. Timely diagnosis and treatment are vital for improving patient outcomes. This study explores the potential role of CXCL5 in the diagnosis, severity assessment, and prognosis of sepsis. We included 147 sepsis patients, 50 patients with systemic inflammatory response syndrome (SIRS) and 120 healthy controls. Serum CXCL5 levels, inflammation scores (APACHE II, SOFA), and other laboratory indicators were recorded. Univariate and multivariate logistic regression analyses were conducted to assess the relationship between CXCL5 and sepsis diagnosis, severity, and prognosis. A prognostic nomogram was constructed and evaluated using receiver operator characteristic curves, calibration curves, and clinical decision curves. Serum CXCL5 levels in sepsis patients were significantly higher than those in patients with SIRS and healthy controls. CXCL5 was identified as a risk factor for sepsis diagnosis. CXCL5 levels were significantly elevated in patients with septic shock (P=0.04) and in deceased patients compared to survivors (P<0.001). The prognostic model, incorporating CXCL5, lactate, APACHE II scores, C-reactive protein levels, and respiratory rate, demonstrated high predictive accuracy with an area under the curve of 0.873. Calibration and decision curve analyses demonstrated the model's good predictive performance and potential clinical value. Serum CXCL5 concentration is a promising biomarker for enhancing the diagnostic accuracy and prognostic evaluation of sepsis. The constructed multivariate prediction model offers new insights into sepsis prognosis, but its direct application in clinical practice requires further validation.

The predictive value of thyroid hormone sensitivity parameters for cervical lymph node metastasis in patients with differentiated thyroid cancer

Objective To comprehensively investigate the predictive value of thyroid hormone sensitivity parameters for cervical lymph node metastasis in patients diagnosed with differentiated thyroid cancer (DTC) undergoing total thyroidectomy and neck lymph node dissection. Methods A retrospective cohort study was conducted involving patients diagnosed with DTC and evaluated for cervical lymph node metastasis. Relevant demographic, tumour, lymph node and thyroid hormone sensitivity parameter data were extracted from medical records and laboratory reports. Thyroid hormone sensitivity parameters including thyroxine (T4), triiodothyronine (T3), thyroid-stimulating hormone (TSH), thyroglobulin (Tg), thyroglobulin antibodies (TgAbs), thyroid peroxidase antibody, thyroid hormone receptor α and TSH receptor antibody were assessed. Statistical analyses including descriptive statistics, comparative analysis, Pearson’s correlation analysis, logistic regression analysis, receiver operating characteristic (ROC) analysis and construction of a multivariable prediction model based on machine learning using the xgbTree method were employed to evaluate the associations and predictive value of thyroid hormone sensitivity parameters for cervical lymph node metastasis. Results The study revealed significant associations between several thyroid hormone sensitivity parameters and cervical lymph node metastasis in patients with DTC. Specifically, higher levels of T4, T3, Tg, TgAbs and TSH receptor antibody were associated with lymph node metastasis. Pearson’s correlation analysis, logistic regression analysis and ROC analysis further underscored the predictive performance of these parameters, with strong overall discriminative abilities. The machine learning-based prediction model demonstrated promising performance with a high area under the curve (AUC) of 0.979. Conclusions The findings provide compelling evidence for the predictive value of thyroid hormone sensitivity parameters, particularly T3, T4, Tg, TgAbs and TSH receptor antibody, in identifying and evaluating the likelihood of cervical lymph node metastasis in patients with DTC. These parameters hold potential implications for risk stratification, clinical decision-making and personalized management strategies, contributing to improved outcomes for patients at risk of lymph node metastasis.