Limitations of Binary Classification for Long-Horizon Diagnosis Prediction and Advantages of a Discrete-Time Time-to-Event Approach: Empirical Analysis.

De Rong Loh^1,2, Elliot D Hill², Nan Liu¹, Geraldine Dawson³, Matthew M Engelhard²

¹Duke-NUS Medical School, 8 College Road, Singapore, 169857, Singapore.

²Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, NC, United States.

³Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, Durham, NC, United States.

PubMed PMC DOI PDF

Abstract

BACKGROUND: A major challenge in using electronic health records (EHR) is the inconsistency of patient follow-up, resulting in right-censored outcomes. This becomes particularly problematic in long-horizon event predictions, such as autism and attention-deficit/hyperactivity disorder (ADHD) diagnoses, where a significant number of patients are lost to follow-up before the outcome can be observed.…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free Account Sign In

Want unlimited research access? See Pro plans

NIH Funding

Grant #11169997
Machine Learning Methods to Develop and Deploy Real-Time Risk Surveillance for Autism Spectrum Disorder and Attention Deficit Hyperactivity Disorder from the Electronic Health Record
DUKE UNIVERSITY
$162,893 (FY2025)
Grant #11393683
Machine Learning Methods to Develop and Deploy Real-Time Risk Surveillance for Autism Spectrum Disorder and Attention Deficit Hyperactivity Disorder from the Electronic Health Record
DUKE UNIVERSITY
$163,204 (FY2026)
Grant #11419214
Duke Autism Center of Excellence: A translational digital health and computational approach to early identification, outcome monitoring, and biomarker discovery in autism
DUKE UNIVERSITY
$2,407,109 (FY2026)

Other Funding

NIMH NIH HHS(K01 MH127309)
United States
NICHD NIH HHS(P50 HD093074)
United States

Keywords

artificial intelligencedeep learningdistributional shiftsearly detectionelectronic health recordsmachine learningpractical modelspredictive modelsright-censoringsurvival analysis

Article Info

PMID: 40605770
PMCID: PMC12223692
DOI: 10.2196/62985
Received: June 7, 2024
Accepted: February 23, 2025

How to Cite

Loh R et al. Limitations of Binary Classification for Long-Horizon Diagnosis Prediction and Advantages of a Discrete-Time Time-to-Event Approach: Empirical Analysis. JMIR AI2025;4e62985. doi:10.2196/62985

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.