• HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
Wednesday, November 18, 2026
BIOENGINEER.ORG
No Result
View All Result
  • Login
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
No Result
View All Result
Bioengineer.org
No Result
View All Result
Home NEWS Science News Health

Ulcerative Colitis Trial Finds Endoscopic Features Add No Value to Treatment Personalization

by
November 10, 2026
in Health
Reading Time: 4 mins read
0
Ulcerative Colitis Trial Finds Endoscopic Features Add No Value to Treatment Personalization

Ulcerative Colitis Trial Finds Endoscopic Features Add No Value to Treatment Personalization

Share on FacebookShare on TwitterShare on LinkedinShare on RedditShare on Telegram

A causal machine learning evaluation of clinical trial data from patients with ulcerative colitis demonstrated that statistically detectable treatment-effect heterogeneity failed to translate into improved treatment assignment, according to a study published in BMC Medical Research Methodology. Conducted by researchers Cristian Minoccheri, Sophia Tesic, Kayvan Najarian, and Ryan Stidham, the investigation revealed that incorporating week-eight video-derived endoscopic features provided no decision value beyond routinely available clinical markers when guiding individualized therapy. The authors reported that while sophisticated machine learning models can identify subtle statistical differences in how patients respond to interventions, those computational detections do not automatically yield superior clinical decision performance in randomized trial settings.

The study addresses a fundamental challenge in modern evidence-based medicine regarding the interpretation of clinical trial results. Randomized controlled trials are primarily designed and conducted to estimate average treatment effects across broad study populations, but substantial heterogeneity in individual patient response motivates personalized therapeutic strategies that attempt to determine which individuals benefit most from specific interventions. Causal machine learning approaches provide a principled methodological foundation for estimating heterogeneous treatment effects and systematically evaluating individualized treatment policies across patient cohorts. Nevertheless, applying these algorithmic techniques to clinical trial datasets demands strict attention to the critical distinction between detecting statistical heterogeneity and achieving measurable improvements in practical decision performance.

To rigorously test these personalized strategies, the investigators developed a comprehensive causal machine learning framework applicable to randomized controlled trials across diverse therapeutic areas. The methodological workflow integrates four primary analytical components: X-learner estimation, permutation-based feature importance, best linear predictor testing, and doubly robust policy-value evaluation combined with out-of-fold validation. The researchers utilized cross-fitted gradient-boosting models to estimate conditional average treatment effects across patient profiles. Statistical confidence intervals across the analyses were computed using bootstrap resampling techniques to preserve uncertainty estimates.

As a concrete case study for the analytical framework, the investigators examined patient-level data derived from the UNIFI maintenance clinical trial evaluating the therapeutic agent ustekinumab in individuals diagnosed with ulcerative colitis. The analysis focused on the randomized maintenance cohort, which comprised three distinct treatment groups. The cohort included a placebo control arm consisting of 144 patients, a standard-dose regimen of ustekinumab administered every twelve weeks consisting of 143 patients, and a dose-intensified regimen of ustekinumab administered every eight weeks consisting of 274 patients.

Within this randomized cohort, the gradient-boosting models estimated conditional average treatment effects by integrating multiple categories of patient information. The candidate predictors comprised baseline demographic characteristics, prior medication history, week-eight clinical scores, routine laboratory biomarkers, and detailed video-derived endoscopic features. A central objective of the research was to assess whether the inclusion of video-derived endoscopic measurements provided demonstrable added value over routinely collected clinical and demographic markers when predicting differential treatment responses.

The initial analytical results showed that standard clinical and demographic features possessed greater aggregate predictive importance than endoscopic variables when modeling treatment-effect heterogeneity. For the comparison between ustekinumab and placebo, non-endoscopic markers registered an aggregate importance score of 0.74 ± 0.02, whereas endoscopic features registered a score of 0.56 ± 0.01, reflecting a 1.33-fold higher relative importance for routine clinical data. Best linear predictor testing demonstrated a statistically significant contribution from endoscopic features when comparing ustekinumab against placebo, yielding a test statistic of z = 7.89 with a p-value of less than 0.001. However, when evaluating the decision between standard therapy and dose intensification, best linear predictor testing indicated no statistically significant contribution from endoscopic features, producing a test statistic of z = 1.18 and a p-value of 0.50.

When the investigators evaluated the actual decision value of these models using doubly robust policy evaluation, the presence of statistical heterogeneity did not produce meaningful improvements in expected clinical remission. In the comparison between ustekinumab and placebo, incorporating endoscopic features yielded an expected remission difference of only positive 4.2 percentage points, with a 95 percent confidence interval spanning from negative 0.2 to positive 8.6 percentage points, indicating no conclusive benefit. More notably, in the out-of-fold three-arm policy evaluation encompassing placebo, standard dosing, and dose intensification, treatment-assignment performance actually worsened when endoscopic features were included in the decision algorithm, resulting in an expected remission rate of 30.9 percent compared to 37.7 percent when assignments relied solely on non-endoscopic clinical markers.

The investigators concluded that week-eight endoscopic features offered no decision value beyond readily available clinical markers within this ulcerative colitis trial cohort, underscoring that statistical heterogeneity detection must be carefully separated from practical clinical utility. By establishing that algorithmic feature importance and statistical significance do not guarantee better patient assignment policies, the framework provides a reproducible template for rigorous causal analysis of treatment personalization in randomized controlled trials across therapeutic disciplines.

Subject of Research: Causal machine learning for treatment personalization in ulcerative colitis trials

Article Title: A causal machine learning framework for treatment personalization in clinical trials: application to ulcerative colitis

Article References: Minoccheri, C., Tesic, S., Najarian, K., & Stidham, R. (2026). A causal machine learning framework for treatment personalization in clinical trials: application to ulcerative colitis. BMC Medical Research Methodology. https://doi.org/10.1186/s12874-026-02982-2

Image Credits: AI Generated

DOI: 10.1186/s12874-026-02982-2

Keywords: causal machine learning, ulcerative colitis, personalized medicine, clinical trials, ustekinumab, heterogeneous treatment effects, endoscopy, treatment policy evaluation, causal, machine, learning, framework

News Source: Ophelia Keating. (November 10, 2026). Ulcerative Colitis Trial Finds Endoscopic Features Add No Value to Treatment Personalization. Scienmag.

Tags: causalcausal machine learningClinical Trialsendoscopyframeworkheterogeneous treatment effectslearningmachinepersonalized medicinetreatment policy evaluationulcerative colitisustekinumab
Share12Tweet7Share2ShareShareShare1

Related Posts

Systematic Review Finds Adolescent Malnutrition Disparities Widening Across Global Socioeconomic Gradients

Systematic Review Finds Adolescent Malnutrition Disparities Widening Across Global Socioeconomic Gradients

November 18, 2026
Deep Learning Recognizes Surgical Phases to Assess Skill in Laparoscopic Procedures

Deep Learning Recognizes Surgical Phases to Assess Skill in Laparoscopic Procedures

November 18, 2026

Sutter Health Launches Registry for Alzheimer-Like Dementia Called LATE

November 18, 2026

Rare NUTM1 Rearranged Sarcoma of the Colon Recurs Aggressively Following Resection

November 18, 2026

POPULAR NEWS

  • Copper Nanoparticles Enhance Corrosion Resistance in Direct-to-Metal Coatings

    Copper Nanoparticles Enhance Corrosion Resistance in Direct-to-Metal Coatings

    29 shares
    Share 12 Tweet 7
  • Electroacupuncture Modulates Hippocampal RNA Fusion Transcripts in Traumatic Brain Injury Rats

    29 shares
    Share 12 Tweet 7
  • Combined Malnutrition and Poor Quality of Life Elevate Digestive Cancer Mortality

    29 shares
    Share 12 Tweet 7
  • Meta-Learning Accelerates Microwave Hyperthermia Antenna Optimization in Proof-of-Concept Study

    29 shares
    Share 12 Tweet 7

About

We bring you the latest biotechnology news from best research centers and universities around the world. Check our website.

Follow us

Recent News

Dual-Stream Swin Transformer Reaches 96 Percent Accuracy Detecting Student Emotions

Systematic Review Finds Adolescent Malnutrition Disparities Widening Across Global Socioeconomic Gradients

Precast Concrete Panels Restore Seismic Resilience in Corroded Underground Infrastructure

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 88 other subscribers
  • Contact Us

Bioengineer.org © Copyright 2023 All Rights Reserved.

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result
  • Homepages
    • Home Page 1
    • Home Page 2
  • News
  • National
  • Business
  • Health
  • Lifestyle
  • Science

Bioengineer.org © Copyright 2023 All Rights Reserved.