{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,12]],"date-time":"2026-10-12T14:44:37Z","timestamp":1791816277436,"version":"4.3.4"},"reference-count":36,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T00:00:00Z","timestamp":1789516800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Public Health"],"abstract":"<jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>To evaluate the predictive information contained in the restricted surveillance feature set (demographic, temporal and specimen variables) for respiratory pathogen identification, and to examine factors associated with model performance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We retrospectively analysed 24,689 acute respiratory infection cases from seven sentinel hospitals in Sichuan, China; after quality control, 21,395 samples with complete 21-pathogen testing were included. Per-pathogen multi-label binary classifiers were trained using 15 features available at presentation; logistic regression, random forest, back-propagation neural network and XGBoost were compared under an identical split and threshold-selection protocol, with thresholds derived from out-of-fold probabilities. Label-definition sensitivity, leave-one-hospital-out cross-validation and prevalence-performance association were analysed.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      Random forest and XGBoost performed comparably (XGBoost the primary model), with Macro-F1 0.1553 (95% CI 0.104\u20130.2091), 0.2202 for common pathogens and Macro-AUC 0.7702; the four-algorithm spread was 0.0611. Log-prevalence correlated strongly with per-pathogen F1 (\n                      <jats:italic>r<\/jats:italic>\n                      \u202f=\u202f0.851); the seven rare pathogens reached Macro-F1 0.0255. Leave-one-hospital-out validation gave Macro-F1 0.1073\u202f\u00b1\u202f0.0289 (\u221230.9%) and Macro-AUC 0.6841 (\u221211.2%). Feature ablation showed temporal and non-temporal features were largely redundant (Macro-F1 0.1153 vs. 0.1147; AUC 0.6892\/0.6865 vs. 0.7702). Treating untested as negative lowered Macro-AUC to 0.7653, whereas simulated false-negative label contamination had minimal effect.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>Models built on the restricted surveillance feature set showed limited diagnostic performance and were not demonstrated suitable for clinical deployment. Performance was strongly associated with pathogen prevalence and constrained by limited feature informativeness, while algorithm choice contributed little and cross-centre generalisability was limited. Priorities include multimodal feature integration, rare-pathogen accrual and prospective multicentre external validation.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fpubh.2026.1961816","type":"journal-article","created":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T05:30:42Z","timestamp":1789536642000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Pathogen prevalence, feature composition and cross-centre generalisability of machine learning diagnostic models for multi-pathogen respiratory infection"],"prefix":"10.3389","volume":"14","author":[{"given":"Fengmiao","family":"Hu","sequence":"first","affiliation":[{"name":"Sichuan Center for Disease Control and Prevention","place":["Chengdu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xingyu","family":"Zhou","sequence":"additional","affiliation":[{"name":"Sichuan Center for Disease Control and Prevention","place":["Chengdu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lijun","family":"Zhou","sequence":"additional","affiliation":[{"name":"Sichuan Center for Disease Control and Prevention","place":["Chengdu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhirui","family":"Li","sequence":"additional","affiliation":[{"name":"Sichuan Center for Disease Control and Prevention","place":["Chengdu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shuang","family":"Dong","sequence":"additional","affiliation":[{"name":"Sichuan Center for Disease Control and Prevention","place":["Chengdu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chongkun","family":"Xiao","sequence":"additional","affiliation":[{"name":"Sichuan Center for Disease Control and Prevention","place":["Chengdu, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,16]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"964","DOI":"10.12138\/j.issn.1671-9638.20233547","article-title":"Construction of a detection and prediction model for suspected respiratory infectious diseases based on hospital clinical data center","volume":"22","author":"Chen","year":"2023","journal-title":"Chin J Infect Control"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"1663234","DOI":"10.3389\/fcimb.2025.1663234","article-title":"tNGS-based detection of respiratory pathogens in a single center: associations with age, gender, season, and co-infections","volume":"15","author":"Wang","year":"2025","journal-title":"Front Cell Infect Microbiol"},{"key":"ref3","doi-asserted-by":"publisher","first-page":"810","DOI":"10.3969\/j.issn.1672-8467.2021.06.014","article-title":"Research on a child influenza a and B auxiliary diagnosis model based on artificial intelligence","volume":"48","author":"Ge","year":"2021","journal-title":"Fudan Univ J Med Sci"},{"key":"ref4","doi-asserted-by":"publisher","first-page":"2954","DOI":"10.11816\/cn.ni.2025-250380","article-title":"Diagnostic model for severe pneumonia caused by adenovirus infection in children based on machine learning and SHAP","volume":"35","author":"Mao","year":"2025","journal-title":"Chin J Nosocomiol"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"2818","DOI":"10.1038\/s41746-026-02818-9","article-title":"Development and validation of a machine learning-based diagnostic system for 22 pediatric respiratory pathogens: a large-scale multicenter study","volume":"9","author":"Su","year":"2026","journal-title":"npj Digital Med"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"e15552","DOI":"10.7717\/peerj.15552","article-title":"Comparative analysis of machine learning approaches for predicting respiratory virus infection and symptom severity","volume":"11","author":"I\u015f\u0131k","year":"2023","journal-title":"PeerJ"},{"key":"ref7","doi-asserted-by":"publisher","first-page":"18029","DOI":"10.1038\/s41598-025-02962-4","article-title":"Predicting and interpreting key features of refractory Mycoplasma pneumoniae pneumonia using multiple machine learning methods","volume":"15","author":"Jiang","year":"2025","journal-title":"Sci Rep"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"1692","DOI":"10.11816\/cn.ni.2026-252930","article-title":"Construction and validation of a machine learning-based model for the differential diagnosis of smear-negative pulmonary tuberculosis and bacterial pneumonia","volume":"36","author":"Cai","year":"2026","journal-title":"Chin J Nosocomiol"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"194","DOI":"10.3760\/cma.j.cn101070-20240822-00527","article-title":"Construction of a machine learning model based on serum indicators and lung ultrasound features for predicting respiratory failure in children with severe pneumonia","volume":"40","author":"Zhou","year":"2025","journal-title":"Chin J Appl Clin Pediatr"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"323","DOI":"10.15953\/j.ctta.2023.066","article-title":"Application of a machine learning fusion model based on CT radiomics for the differential diagnosis of COVID-19 and other viral pneumonia in children","volume":"32","author":"Li","year":"2023","journal-title":"CT Theory Applications"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"122","DOI":"10.3969\/j.issn.1672-9528.2025.08.030","article-title":"Research on respiratory disease diagnosis models based on machine learning","volume":"8","author":"Zhang","year":"2025","journal-title":"Info Technol Informatization"},{"key":"ref12","doi-asserted-by":"publisher","first-page":"55","DOI":"10.7326\/M14-0697","article-title":"Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement","volume":"162","author":"Collins","year":"2015","journal-title":"Ann Intern Med"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"1","DOI":"10.18637\/jss.v045.i03","article-title":"Mice: multivariate imputation by chained equations in R","volume":"45","author":"van Buuren","year":"2011","journal-title":"J Stat Softw"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"44","DOI":"10.1093\/nsr\/nwx106","article-title":"A brief introduction to weakly supervised learning","volume":"5","author":"Zhou","year":"2018","journal-title":"Natl Sci Rev"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"673","DOI":"10.1007\/s11135-006-9018-6","article-title":"A caution regarding rules of thumb for variance inflation factors","volume":"41","author":"O\u2019Brien","year":"2007","journal-title":"Qual Quant"},{"key":"ref16","author":"Chen","year":"2016"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"321","DOI":"10.1613\/jair.953","article-title":"SMOTE: synthetic minority over-sampling technique","volume":"16","author":"Chawla","year":"2002","journal-title":"J Artif Intell Res"},{"key":"ref18","author":"Lin","year":"2017"},{"key":"ref19","first-page":"3146","article-title":"LightGBM: a highly efficient gradient boosting decision tree","volume":"30","author":"Ke","year":"2017","journal-title":"Adv Neural Inf Proces Syst"},{"key":"ref20","first-page":"6638","article-title":"CatBoost: unbiased boosting with categorical features","volume":"31","author":"Prokhorenkova","year":"2018","journal-title":"Adv Neural Inf Proces Syst"},{"key":"ref21","author":"Arik","year":"2021"},{"key":"ref22","doi-asserted-by":"publisher","first-page":"91","DOI":"10.1186\/1471-2105-7-91","article-title":"Bias in error estimation when using cross-validation for model selection","volume":"7","author":"Varma","year":"2006","journal-title":"BMC Bioinformatics"},{"key":"ref23","doi-asserted-by":"publisher","first-page":"2079","DOI":"10.5555\/1756006.1859921","article-title":"On over-fitting in model selection and subsequent selection bias in performance evaluation","volume":"11","author":"Cawley","year":"2010","journal-title":"J Mach Learn Res"},{"key":"ref24","doi-asserted-by":"publisher","first-page":"3083","DOI":"10.1002\/sim.3697","article-title":"Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples","volume":"28","author":"Austin","year":"2009","journal-title":"Stat Med"},{"key":"ref25","doi-asserted-by":"publisher","first-page":"1263","DOI":"10.1109\/TKDE.2008.239","article-title":"Learning from imbalanced data","volume":"21","author":"He","year":"2009","journal-title":"IEEE Trans Knowl Data Eng"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"1340","DOI":"10.1093\/bioinformatics\/btq134","article-title":"Permutation importance: a corrected feature importance measure","volume":"26","author":"Altmann","year":"2010","journal-title":"Bioinformatics"},{"key":"ref27","first-page":"4765","article-title":"A unified approach to interpreting model predictions","volume":"30","author":"Lundberg","year":"2017","journal-title":"Adv Neural Inf Proces Syst"},{"key":"ref28","doi-asserted-by":"publisher","first-page":"230","DOI":"10.1186\/s12916-019-1466-7","article-title":"Calibration: the Achilles heel of predictive analytics","volume":"17","author":"Van Calster","year":"2019","journal-title":"BMC Med"},{"key":"ref29","doi-asserted-by":"publisher","first-page":"128","DOI":"10.1097\/EDE.0b013e3181c30fb2","article-title":"Assessing the performance of prediction models: a framework for traditional and novel measures","volume":"21","author":"Steyerberg","year":"2010","journal-title":"Epidemiology"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"1682764","DOI":"10.3389\/fcimb.2025.1682764","article-title":"An interpretable machine learning model for early prediction of Escherichia coli infection in ICU patients","volume":"15","author":"Yang","year":"2025","journal-title":"Front Cell Infect Microbiol"},{"key":"ref31","doi-asserted-by":"publisher","first-page":"261","DOI":"10.1038\/s43856-025-00965-w","article-title":"Diagnostic framework to validate clinical machine learning models locally on temporally stamped data","volume":"5","author":"Schuessler","year":"2025","journal-title":"Commun Med"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"e78309","DOI":"10.2196\/78309","article-title":"Unsupervised characterization of temporal dataset shifts as an early indicator of AI performance variations: evaluation using MIMIC-IV","volume":"13","author":"Fern\u00e1ndez-Narro","year":"2025","journal-title":"JMIR Med Inform"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1101\/2025.07.31.25332542","article-title":"Evaluating deep learning sepsis prediction models under distribution shift: a multi-Centre retrospective cohort study","author":"Tranchellini","year":"2025","journal-title":"medRxiv"},{"key":"ref34","doi-asserted-by":"publisher","first-page":"e0001534","DOI":"10.1371\/journal.pdig.0001534","article-title":"Validation is not enough: longitudinal evidence of post-deployment fragility in clinical AI systems","volume":"5","author":"Kopanitsa","year":"2026","journal-title":"PLOS Digital Health"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"51","DOI":"10.7326\/M18-1376","article-title":"PROBAST: a tool to assess the risk of bias and applicability of prediction model studies","volume":"170","author":"Wolff","year":"2019","journal-title":"Ann Intern Med"},{"key":"ref36","doi-asserted-by":"publisher","first-page":"565","DOI":"10.1177\/0272989X06295361","article-title":"Decision curve analysis: a novel method for evaluating prediction models","volume":"26","author":"Vickers","year":"2006","journal-title":"Med Decis Mak"}],"updated-by":[{"DOI":"10.3389\/fpubh.2026.2015265","type":"correction","label":"Correction","source":"publisher","updated":{"date-parts":[[2026,10,12]],"date-time":"2026-10-12T00:00:00Z","timestamp":1791763200000}}],"container-title":["Frontiers in Public Health"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpubh.2026.1961816\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,12]],"date-time":"2026-10-12T13:45:22Z","timestamp":1791812722000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpubh.2026.1961816\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,16]]},"references-count":36,"alternative-id":["10.3389\/fpubh.2026.1961816"],"URL":"https:\/\/doi.org\/10.3389\/fpubh.2026.1961816","relation":{},"ISSN":["2296-2565"],"issn-type":[{"value":"2296-2565","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,16]]},"article-number":"1961816"}}