This observational study aims to (1) validate a multimodal artificial intelligence (AI) model for early detection of cancer-associated cachexia in pancreatic cancer patients and (2) assess the feasibility and acceptability of diet and exercise interventions for cachexia management. The study will use retrospective data from the Florida Pancreas Collaborative and prospective data from newly diagnosed patients at Moffitt Cancer Center.

Inclusion Criteria:

• Age ≥ 18 years
• Histologically or cytologically confirmed diagnosis of pancreatic exocrine malignancy (e.g.pancreatic ductal adenocarcinoma, acinar cell carcinoma, pancreatic adenosquamous carcinoma, etc.) or clinical features highly suggestive of such malignancy with diagnostic confirmation anticipated during the screening or early study period.
• Presentation consistent with resectable, borderline resectable, local, or locally advanced, or metastatic disease
• May or may not have had surgery
• Treatment plan likely includes systemic therapy for pancreatic cancer
• ECOG performance status 0-2
• Able to tolerate oral intake and not currently receiving enteral or parenteral nutrition
• Able to read and speak English
• Able to provide written informed consent

Exclusion Criteria:

• Pregnant or breastfeeding at the time of enrollment
• Current use of tube feeding (enteral) or total parenteral nutrition
• Presence of ascites or other findings suggestive of decompensated disease
• Evidence or history of bowel obstruction or any gastrointestinal condition that may limit food intake
• Severe or uncontrolled psychiatric illness (e.g., psychosis, dementia) that would interfere with participation
• Presence of other uncontrolled intercurrent illnesses (e.g., infection, heart failure, liver failure) that may interfere with protocol adherence

Adults (≥18 years) with histologically or cytologically confirmed pancreatic ductal adenocarcinoma (PDAC), presenting with resectable, borderline resectable, or locally advanced nonmetastatic disease. ECOG performance status 0-2. Total enrollment includes 100 newly diagnosed pancreatic cancer patients (cross-sectional), a longitudinal subset of 50 patients followed for 9 months, and 20 oncology clinicians completing feasibility surveys.