AI-Driven Tumor Response Evaluation for Solid Tumors

Development of an Artificial Intelligence-Driven Novel Response Evaluation Framework and Its Biological Characterization

Purpose: This study is developing and validating an artificial intelligence (AI)-driven system to evaluate tumor response using changes in total tumor volume. The goal is to determine whether this AI-based approach can better predict patient survival compared with the current standard method (RECIST), which relies on linear measurements of a few selected tumors.

Participants: The study includes both retrospective and prospective cohorts. The retrospective cohort includes approximately 6,000 patients with solid tumors who received non-surgical treatment between 2015 and 2025. The prospective cohort will enroll approximately 120 patients starting in mid-2026.

Study details include:

Study Duration: Approximately 3 years

Participation Duration: Up to 6 months for prospective participants; retrospective participants contribute existing medical records only

Visit Frequency: For prospective participants, follow-up visits occur every 3 months (up to 6 months) aligned with routine clinical care

Intervention: None. This is an observational study using routine clinical imaging (CT/MRI) and medical records

Primary endpoints: Overall survival (OS) and progression-free survival (PFS). The study will also evaluate the feasibility and impact of AI-assisted tumor response reporting on clinical workflow and patient understanding.

Participants in the prospective cohort will receive either a standard RECIST report or an AI-assisted dynamic tumor response report. This comparison is for research purposes only and does not alter standard medical care.

Study Overview

• Status: Not yet recruiting
• Conditions: Neoplasms; Hepatocellular Carcinoma; Solid Tumor Cancer
• Intervention / Treatment: Other: Observational Data Collection

Detailed Description

Background: Current tumor response evaluation relies primarily on RECIST 1.1 and its variants, which measure changes in the longest diameter of a limited number of target lesions. While standardized and widely used, these criteria have limitations: they may not fully reflect total tumor burden changes, fail to capture spatial and temporal heterogeneity across lesions, and are subject to inter-observer variability. Advances in artificial intelligence, particularly in medical image analysis, now enable automated tumor segmentation and volumetric quantification, offering a more comprehensive assessment of tumor burden dynamics. However, systematic validation of AI-driven volume-based response criteria against traditional methods remains limited.

Study Design: This is a multi-center, retrospective-prospective cohort study designed to develop and validate an AI-driven prognostic model based on total tumor volume changes and multi-dimensional features. The study is being conducted across approximately 40 participating sites in China.

Data Sources:

Training Set (Retrospective): Approximately 6,000 patients with solid tumors who received non-surgical treatment between January 2015 and December 2025, including the Hepatorch cohort (~300 patients). Data include imaging (CT/MRI), clinical characteristics, laboratory tests, and molecular markers.

Validation Set: Approximately 20% of the training set data randomly extracted for hyperparameter tuning and internal validation.

External Test Set (Prospective): Approximately 120 patients consecutively enrolled from mid-2026 through 2027, independent of the training/validation sets, to assess model generalizability in real-world clinical settings.

AI Model Development: The prediction model integrates imaging biomarkers (total tumor volume, single-lesion volume, enhanced volume, lesion count), clinical variables (age, sex, tumor type/stage, liver function, treatment modality), and laboratory/molecular markers (AFP, CA19-9, immune markers, genetic sequencing). Modeling approaches include joint models for longitudinal volume changes and survival outcomes, random survival forests, gradient boosting survival models, and deep learning survival models. Model performance is evaluated using C-index, time-dependent ROC curves, calibration curves, and decision curve analysis.

Prospective Sub-studies:

Lesion Tracking and Biological Characterization (~20 patients): Serial tracking of individual lesions with volumetric measurement, plus collection of leftover tumor tissue from clinically indicated biopsies for molecular and immune microenvironment analysis.

Patient Experience and Communication Value Assessment (~100 patients): Participants are randomized 1:1 to receive either a standard RECIST report or an AI-assisted dynamic tumor response report. Standardized questionnaires assess report comprehension, cognitive burden, anxiety, trust, and treatment decision confidence.

Human-Machine Collaboration: In both retrospective and prospective components, the study evaluates AI-assisted tumor response assessment by comparing independent clinician reading, AI-assisted reading, and expert adjudication. Consistency metrics include Kappa statistics, intraclass correlation coefficients, and measurement error. Efficiency metrics include reading time and report generation time.

Follow-up Schedule: For the prospective cohort, follow-up visits occur every 3 months (up to 6 months) aligned with routine clinical care, collecting imaging, laboratory results, treatment changes, disease progression, survival status, and subsequent therapy information.

Statistical Considerations: Model development employs LASSO regularization, stepwise regression, and machine learning-based feature selection. Model validation uses K-fold cross-validation and independent external validation. Predictive performance is assessed using time-dependent ROC, C-index, calibration plots, and decision curve analysis. Missing data are handled using multiple imputation.

Ethical Considerations: The retrospective component uses de-identified data from patients who previously consented to biobank participation, with a waiver of informed consent requested. The prospective component requires written informed consent from all participants. The study protocol has been approved by the Institutional Review Board of Zhongshan Hospital, Fudan University.

• Study Type: Observational
• Enrollment (Estimated): 6120

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

This multi-center retrospective-prospective cohort study includes adult patients (≥18 years) with radiologically or pathologically confirmed solid tumors who have received or are receiving non-surgical treatment. The retrospective cohort comprises approximately 6,000 patients treated between January 2015 and December 2025, with complete imaging, clinical, laboratory, and molecular data from multiple centers across China. The prospective cohort consists of approximately 120 patients consecutively enrolled from 2026 onward, with data collected in real-world clinical settings. Participants are excluded if imaging quality is insufficient for AI-based analysis, key clinical or follow-up data are missing, or they have concurrent malignancies that cannot be distinguished from the primary tumor. The study population reflects the diversity of solid tumor types and treatment patterns in routine clinical practice.

Description

Inclusion Criteria:

1. Age ≥ 18 years, any sex.
2. Radiologically or pathologically confirmed diagnosis of solid tumor.
3. Received non-surgical treatment with a clearly defined treatment start date.
4. Availability of baseline and at least one follow-up imaging study (CT/MRI) of sufficient quality for AI-based segmentation and volumetric analysis.
5. Availability of key clinical data and follow-up outcome information.
6. For retrospective cohort: prior signed informed consent for biobank donation, agreeing to donate samples and data for medical research.
7. For prospective cohort: planned to receive or currently receiving non-surgical treatment, and able to provide written informed consent.

Exclusion Criteria:

1. Imaging data incomplete or of insufficient quality for accurate segmentation or volumetric calculation.
2. Key clinical information or follow-up outcome data missing.
3. Treatment start or baseline time point cannot be clearly determined.
4. Concurrent other malignancy that cannot be distinguished from the primary study tumor.
5. Severe underlying diseases (e.g., cardiac, pulmonary, renal insufficiency) that may significantly affect survival outcome assessment.
6. Cognitive impairment or other conditions that prevent cooperation with study procedures.
7. For prospective cohort: expected inability to complete follow-up.
8. Other conditions judged by the investigator as unsuitable for study inclusion. -

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
Retrospective Cohort; Approximately 6,000 patients with solid tumors who received non-surgical treatment between January 2015 and December 2025. Data are collected from existing medical records, imaging archives (CT/MRI), and laboratory databases, with no additional interventions or procedures. This cohort is used for AI model training and internal validation.	Other: Observational Data Collection; This is an observational study. No interventions are assigned. Data are collected from routine clinical imaging (CT/MRI), medical records, and laboratory tests as part of standard clinical care.
Prospective Cohort; Approximately 120 patients with solid tumors consecutively enrolled from 2026 onward. Data are collected prospectively in real-world clinical settings using an EDC system, including imaging, clinical, laboratory, and molecular data. Participants undergo standard-of-care imaging and follow-up; no study-specific interventions are assigned. This cohort is used for independent external validation of the AI model and assessment of clinical feasibility and patient experience.	Other: Observational Data Collection; This is an observational study. No interventions are assigned. Data are collected from routine clinical imaging (CT/MRI), medical records, and laboratory tests as part of standard clinical care.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Overall Survival (OS)	Time from treatment initiation to death from any cause or last follow-up. OS is an objective, clinically meaningful endpoint that directly reflects treatment efficacy and patient prognosis.	From treatment initiation until death or last follow-up, assessed up to 36 months

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Progression-Free Survival (PFS)	Time from treatment initiation to first documented disease progression or death from any cause.	From treatment initiation until disease progression or death, assessed up to 36 months
Tumor Volume Change Rate	Percentage change in total tumor volume from baseline, as measured by AI-based automated segmentation on CT/MRI imaging.	Baseline and at each follow-up imaging time point (e.g., 4-8 weeks, 3 months, 6 months, 12 months post-treatment), assessed up to 36 months
Change in Number of Lesions	Change in the total number of tumor lesions from baseline, as identified by AI-based detection on CT/MRI imaging.	Baseline and at each follow-up imaging time point, assessed up to 36 months
Appearance of New Lesions	Presence or absence of new tumor lesions identified on follow-up CT/MRI imaging compared to baseline.	At each follow-up imaging time point, assessed up to 36 months

Other Outcome Measures

Outcome Measure	Measure Description	Time Frame
AI Model Performance - Concordance Index (C-index)	Concordance index of the AI-driven prognostic model for predicting overall survival, evaluated in the prospective validation cohort.	At study completion, assessed up to 36 months
AI Model Performance - Time-Dependent ROC AUC	Time-dependent Area Under the Receiver Operating Characteristic Curve (AUC) of the AI-driven prognostic model for predicting survival at specific time points (e.g., 12, 24 months).	At study completion, assessed up to 36 months
Physician Workflow Efficiency - Reading Time	Time required for radiologists or clinicians to complete tumor response assessment, comparing AI-assisted reading versus independent manual reading.	During the human-machine collaboration evaluation, assessed up to 36 months

Collaborators and Investigators

• Sponsor: Shanghai Zhongshan Hospital
• Collaborators: Fudan University; West China Hospital; Peking University Cancer Hospital & Institute; Sun Yat-Sen University Cancer Center

Study record dates

Study Major Dates

• Study Start (Estimated): July 1, 2026
• Primary Completion (Estimated): June 30, 2029
• Study Completion (Estimated): June 30, 2029

Study Registration Dates

• First Submitted: June 28, 2026
• First Submitted That Met QC Criteria: June 28, 2026
• First Posted (Actual): July 6, 2026

Study Record Updates

• Last Update Posted (Actual): July 8, 2026
• Last Update Submitted That Met QC Criteria: July 6, 2026
• Last Verified: July 1, 2026

More Information

Terms related to this study

• Keywords: Artificial Intelligence; Tumor Response Evaluation
• Additional Relevant MeSH Terms: Neoplasms by Site; Neoplasms by Histologic Type; Digestive System Neoplasms; Digestive System Diseases; Liver Diseases; Neoplasms, Glandular and Epithelial; Adenocarcinoma; Liver Neoplasms; Carcinoma; Neoplasms; Carcinoma, Hepatocellular
• Other Study ID Numbers: AIRECIST-001

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No