A Study on a Predictive Model for Efficacy and Prognosis of Pancreatic Carcinoma Based on Multimodal Data

A Study on a Predictive Model for Efficacy and Prognosis of Pancreatic Carcinoma Based on Multimodal Data

The goal of this observational study is to learn if combining information from CT scans, blood tests, and pathology reports can better predict how pancreatic cancer will progress.

The main questions it aims to answer are:

• Can combining these types of data more accurately estimate how long a person might survive?
• Can it better predict the risk of recurrence?

Participants will not have any extra tests or treatments. They will:

• Allow researchers to collect information from their existing medical records (such as surgery reports, imaging, and lab results)
• Receive a follow-up phone call for up to 3 years to share health updates（about every 3 months for the prospective cohort）.

Inclusion Criteria:

• Age ≥ 18 years.
• Histologically or cytologically confirmed diagnosis of pancreatic cancer.
• Availability of preoperative and postoperative contrast-enhanced CT images at the participating center; patient records include corresponding clinical data, postoperative pathological reports, and preoperative and postoperative blood test results.
• Underwent surgical treatment for pancreatic cancer at a participating center of this study.

Exclusion Criteria:

• Age < 18 years (minors).
• Presence of other severe diseases (e.g., severe liver or kidney failure, cardiovascular or cerebrovascular diseases, malignancies other than pancreatic cancer); pregnant or breastfeeding women; individuals with mental illness;
• patients unable to comply with follow-up or provide informed consent.
• Presence of significant outliers (e.g., laboratory values exceeding 10 times the normal range without clinically reasonable explanation) or samples with excessive missing data.