Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 24, 2026

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

Making LLM Predictions Interpretable: Fine-Tuning GPT-4o for Early Discontinuation of Cancer Medication.

Congning Ni¹, Qingyuan Song¹, Jeremy L Warner²

¹Vanderbilt University, Nashville, TN, USA.

Studies in Health Technology and Informatics

|May 23, 2026

Summary

Related Concept Videos

Combination Therapies and Personalized Medicine

Combination Therapies and Personalized Medicine

Combining two or more treatment methods increases the life span of cancer patients while reducing damage to vital organs or tissue from the overuse of a single treatment. Combination therapy also targets different cancer-inducing pathways, thus reducing the chances of developing resistance to treatment.
The combination of the drug acetazolamide and sulforaphane is a good example of combination therapy to treat cancer. The cells in the interior of a large tumor often die due to the hypoxic and...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Announcing a New Article Type in <i>JCO Clinical Cancer Informatics</i>: The Resource Report.

JCO clinical cancer informatics·2026

Same author

Implementing trust in non-small cell lung cancer diagnosis with a conformalized uncertainty-aware AI framework.

Nature biomedical engineering·2026

Same author

Generating synthetic multi-national longitudinal cohorts for clinically grounded HIV research.

Nature communications·2026

Same author

Multi-Modal Deep Learning-Based Model to Predict Burkitt Lymphoma Recurrence.

AMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science·2026

Same author

A real-world feasibility evaluation of LLM-based clinical prediction: emergency department return visit admission across two academic medical centers.

Research square·2026

Same author

High safety screening rates prior to immunomodulator initiation at health-system specialty pharmacies: A multisite, prospective, observational cohort study.

American journal of health-system pharmacy : AJHP : official journal of the American Society of Health-System Pharmacists·2026

Predicting medication discontinuation in cancer patients is crucial. A fine-tuned GPT-4o model outperformed traditional machine learning, offering a promising tool for early risk detection and enhanced clinical interpretability.

Area of Science:

Oncology
Artificial Intelligence
Clinical Informatics

Background:

Medication discontinuation presents a significant challenge in cancer care, impacting treatment efficacy.
Early prediction of premature treatment cessation is vital for proactive patient management and care team awareness.

Purpose of the Study:

To compare the performance of a general-purpose large language model (LLM), GPT-4o, against traditional machine learning (ML) models for predicting medication discontinuation using structured electronic health record (EHR) data.
To assess the capability of LLMs to generate clinician-readable rationales for predictions, facilitating interpretability and comparison with feature attribution methods.

Main Methods:

A comparative analysis was conducted using 2,364 patient records from a major academic medical center.

Keywords:

GPT-4o large language models medication discontinuation prediction

Related Experiment Videos

Last Updated: May 24, 2026

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma

Published on: October 10, 2018

A fine-tuned GPT-4o model was evaluated against traditional ML models, including XGBoost.

Interpretability was assessed using SHAP values for XGBoost and a SHAP-like mimic attribution for GPT-4o.

Main Results:

Fine-tuned GPT-4o achieved a superior F1 score of 0.867, outperforming the best traditional model, XGBoost (F1=0.825).
Both SHAP (XGBoost) and the GPT-4o mimic attribution prioritized similar clinical features, such as age and BMI, demonstrating aligned interpretability.
LLMs demonstrated competitive performance in structured EHR prediction tasks.

Conclusions:

Large language models, specifically fine-tuned GPT-4o, show significant promise for structured clinical prediction tasks in oncology.
LLMs offer a dual benefit of high predictive accuracy and inherent interpretability, enabling direct comparison with established ML interpretability techniques.
These findings support the integration of LLMs into clinical workflows for improved patient monitoring and risk stratification.