Pancreatic cancer search behavior and early detection signals

The deadliest cancer is a data problem

Pancreatic cancer: stage at initial diagnosis (SEER, 2023)

Pancreatic cancer has a five-year survival rate of just 13%. Not because treatments do not exist, but because 80% of cases are diagnosed at stage III or IV, when options are limited and outcomes are grim. The pancreas sits deep in the abdomen, produces vague symptoms, and evades standard screening protocols.

The result: by the time a patient presents with jaundice, unexplained weight loss, or persistent back pain, the window for curative surgery has usually closed.

This is not solely a clinical failure. It is an intelligence failure. Pancreatic cancer data exists across fragmented systems: lab records, pharmacy claims, search behavior, wearable outputs, and primary care notes. The signals are there. They are just not being connected, scored, or trusted.

What behavioral signals actually precede a pancreatic cancer diagnosis?

Researchers at institutions including Harvard and the Mayo Clinic have identified clusters of pre-diagnostic behavioral and clinical changes that appear months before a formal pancreatic cancer diagnosis.

These include:

New-onset diabetes in adults over 50. Roughly 1% of new diabetes cases in this age group are caused by pancreatic tumors. A 2023 study in Gastroenterology found that sudden glycemic changes 6 to 18 months before diagnosis were a statistically significant signal.

Unexplained weight loss paired with gastrointestinal complaints, especially in patients with no prior GI history.

Changes in prescription patterns. Patients frequently receive new prescriptions for PPIs, pancreatic enzymes, or anti-nausea medications in the 12 months before diagnosis.

Search behavior shifts. People search for symptoms like "why does my back hurt after eating" or "sudden blood sugar spike causes" long before they search for "pancreatic cancer symptoms."

These are pancreatic cancer early detection behavioral signals. Individually, each one is noise. Together, scored against a validated framework, they become actionable intelligence.

Does pancreatic cancer show in blood tests?

Yes, but imperfectly. The most commonly referenced biomarker is CA 19-9, a protein that is elevated in many pancreatic cancer patients. However, CA 19-9 is not reliable for early detection. It produces false positives in patients with bile duct obstruction, pancreatitis, and other non-cancerous conditions. It also misses roughly 10% of the population that does not produce the antigen at all.

Newer liquid biopsy approaches are more promising. A multi-cancer early detection blood test from Grail (Galleri) can identify pancreatic cancer signals via cell-free DNA, though sensitivity for stage I pancreatic cancer remains below 50%. CancerSEEK, developed at Johns Hopkins, combines protein biomarkers with ctDNA and showed 72% sensitivity in a 2018 Science study.

No single blood test is the most accurate test for pancreatic cancer. The future is multi-modal: blood biomarkers, imaging, behavioral data, and clinical history combined and weighted by reliability.

Why pancreatic cancer intelligence depends on data trust

Here is the core problem. Behavioral signals, lab results, claims data, and genomic markers all come from different sources with different collection standards, consent models, and quality levels. If you build a detection model on unreliable data, you get unreliable predictions.

This is exactly what the Data Trust Index (DTI) was designed to solve. Every health data record scored 0 to 100 across eight dimensions: provenance, consent, recency, quality, concordance, validation, breadth, and stability. Think of it as a FICO score for health data.

When SuperTruth worked with imaware, a diagnostics company, we standardized 105,000 records, reducing processing time from three weeks to two hours and identifying a patient segment driving 20% of revenue. That same rigor applies to pancreatic cancer data pipelines.

VIOLET, our oncology behavioral intelligence product, is built to surface exactly the kinds of signals described above. Not by replacing clinicians, but by giving them scored, contextualized pancreatic cancer intelligence before symptoms become emergencies.

How to detect pancreatic cancer symptoms before they escalate

You cannot test for pancreatic cancer at home with any validated consumer product today. But you can be aware of the combination of signals that warrant further evaluation: new-onset diabetes after 50, unintentional weight loss exceeding 5% of body weight, persistent upper abdominal or mid-back pain, and digestive changes without a clear cause.

The clinical path forward involves EUS (endoscopic ultrasound) and MRI/MRCP for high-risk individuals, particularly those with family history or known genetic mutations like BRCA2, PALB2, or Lynch syndrome.

The data path forward is just as critical. We need systems that can ingest, score, and connect fragmented behavioral and clinical signals at scale, with consent and provenance baked in from the start.

What comes next

Pancreatic cancer will not be solved by a single breakthrough test. It will be solved by connecting trusted data across sources, scoring it rigorously, and putting behavioral intelligence into the hands of clinicians and researchers who can act on it.

SuperTruth is building that infrastructure now.

Further reading: See VIOLET: behavioral signal intelligence for oncology To learn how VIOLET and the DTI Engine can support early detection research and oncology data strategies, contact Louis Simeonidis, SVP of Commercial Operations, at louis@supertruth.ai or (215) 918-4140.