AI Multimodal Model for Predicting CRPC Progression Risk

Artificial Intelligence for Predicting Progression Risk of Castration-Resistant Prostate Cancer by Integrating Multimodal Data: A Multicenter, Retrospective Study

Status

Active, not recruiting

Phases

Unknown

Study type

Observational

Source

ClinicalTrials.gov

Registry ID

NCT07658573

Acronym

AI-MM_CPRC

Enrollment

500

Registered

2026-06-22

Start date

2025-06-01

Completion date

2028-06-01

Last updated

2026-06-22

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Castration-Resistant Prostate Cancer, Prostate Cancer

Brief summary

This study aims to develop an artificial intelligence model to predict which patients with advanced prostate cancer are at higher risk of developing castration-resistant prostate cancer (CRPC), a more severe form of the disease. The study will use pre-treatment MRI images, biopsy pathology slides, and clinical data collected from patients who received either hormone therapy (ADT) or radical prostatectomy surgery. By integrating these different types of data, the AI model is designed to help doctors identify high-risk patients earlier, personalize treatment plans, and ultimately improve patient outcomes. This is a multicenter, retrospective study that will analyze data from over 500 patients with at least 24 months of follow-up. The performance of the model will be evaluated using standard accuracy metrics.

Interventions

None listed

Study design

Observational model

COHORT

Time perspective

PROSPECTIVE

Eligibility

Sex/Gender

MALE

Age

18 Years to 80 Years

Healthy volunteers

Inclusion criteria

* (1) Male, age ≥18 years; (2) Pathologically confirmed newly diagnosed advanced prostate cancer; (3) Received at least 6 months of androgen deprivation therapy OR underwent radical prostatectomy; (4) Have complete baseline MRI (T2WI, DWI, ADC) and biopsy pathology data; (5) Have complete follow-up records (at least 24 months).

Exclusion criteria

* (1) Missing clinical information; (2) MRI images of poor quality or missing key sequences; (3) Poor quality pathological specimens; (4) Lost to follow-up or incomplete data during follow-up; (5) Concurrent other malignant tumors; (6) Prior anti-tumor therapy.

Design outcomes

Primary

Measure	Time frame	Description
Prediction of CRPC Progression Risk in the ADT Treatment Group	Minimum 24 months of follow-up	Performance of the multimodal AI model in predicting progression to castration-resistant prostate cancer in patients with newly diagnosed advanced prostate cancer who received at least 6 months of androgen deprivation therapy. Metrics include AUC, accuracy, sensitivity, and specificity.
Prediction of CRPC Progression Risk in the Radical Prostatectomy Group	Minimum 24 months of follow-up	Performance of the multimodal AI model in predicting progression to castration-resistant prostate cancer in patients with newly diagnosed advanced prostate cancer who underwent radical prostatectomy. Metrics include AUC, accuracy, sensitivity, and specificity.

Countries

China

Outcome results