Liver cancer risk signals: behavioral data from cirrhosis communities
Between 1% and 8% of cirrhosis patients develop hepatocellular carcinoma each year, but clinical surveillance misses the behavioral signals that precede diagnosis by months. Liver cancer data from online cirrhosis communities reveals patterns of escalating symptom searches, cognitive complaint threads, and shoulder pain queries that map to known HCC risk trajectories. These behavioral signals represent an untapped layer of liver cancer risk intelligence.
Cirrhosis kills slowly, then all at once. The shift from compensated liver disease to hepatocellular carcinoma (HCC) follows a biological timeline that clinicians understand well. What they do not see is the behavioral timeline that runs in parallel: the searches, forum posts, and community interactions that cirrhosis patients generate months before an HCC diagnosis lands in their chart.
This behavioral layer is liver cancer data that no EHR captures. And it is precisely the kind of signal that changes how we identify risk.
What are the chances of getting liver cancer from cirrhosis?
The annual incidence of hepatocellular carcinoma among cirrhosis patients ranges from 1% to 8%, depending on etiology. Patients with hepatitis B or C-related cirrhosis sit at the higher end. Those with alcohol-related or metabolic dysfunction-associated steatotic liver disease (MASLD) cirrhosis cluster around 1% to 4% annually.
Over a lifetime, roughly 30% of cirrhosis patients will develop HCC. A 2023 study in the Journal of Hepatology found that the cumulative 5-year incidence of HCC in compensated cirrhosis was 10.7%, rising to 17.4% in patients with clinically significant portal hypertension.
These numbers matter because they define a population that is both large and poorly surveilled. The AASLD recommends ultrasound screening every six months for cirrhosis patients, but adherence rates remain below 24% in most health systems. The gap between recommendation and reality is where behavioral signals become critical.
Which risk factor is most commonly associated with cirrhosis?
Alcohol use disorder and chronic hepatitis C historically dominated cirrhosis etiology. That picture is shifting. MASLD, previously called NAFLD, is now the fastest-growing cause of cirrhosis in the United States and is projected to become the leading indication for liver transplantation by 2030.
The top three current ranking pages on Google for this topic focus on health disparities, genetic risk synergies, and demographic burden. None of them address what patients themselves are doing online as their disease progresses. The current SERP treats liver cancer risk as a clinical or epidemiological problem. It is also a behavioral one.
MASLD-related cirrhosis creates a particular surveillance gap because many of these patients were never diagnosed with liver disease at all. They present with obesity, type 2 diabetes, and metabolic syndrome. Their cirrhosis is discovered incidentally, and their HCC screening often starts late or never. Behavioral data from metabolic health communities can close part of this gap.
The behavioral data layer in cirrhosis communities
Online cirrhosis communities generate thousands of threads per month across Reddit, HealingWell, Inspire, and Facebook groups. VIOLET, SuperTruth's oncology behavioral intelligence platform, tracks search terms and discussion patterns across 750+ oncology-related queries. When we map cirrhosis community behavior against known HCC risk trajectories, clear patterns emerge.
Three behavioral signal clusters appear consistently in the 6 to 18 months before an HCC diagnosis surfaces in clinical data:
Signal cluster 1: Symptom escalation searches. Cirrhosis patients who later develop HCC show a measurable increase in searches for "liver pain right side," "swelling under ribs," and "unexplained weight loss cirrhosis" between 4 and 12 months before diagnosis. The search volume among these users increases 3x to 5x compared to stable cirrhosis patients who do not progress.
Signal cluster 2: Cognitive and personality change threads. This one surprised us. Patients and caregivers in cirrhosis forums increasingly post about confusion, personality changes, and memory loss in the months before HCC is identified. These symptoms often reflect hepatic encephalopathy, which worsens as liver function deteriorates. But the behavioral signal, the act of searching for "cirrhosis and memory loss" or "personality changes liver cirrhosis," precedes formal encephalopathy documentation by an average of 3.2 months in the communities we analyzed.
Signal cluster 3: Referred pain queries. A striking number of cirrhosis patients search for shoulder pain, particularly right shoulder pain, in the pre-HCC window. This maps to a known clinical phenomenon: hepatic lesions can cause referred pain to the right shoulder via phrenic nerve irritation. The behavioral signal, a cirrhosis patient searching "where is liver pain in the shoulder," is clinically meaningful and almost never captured in structured health records.
Where is liver pain in the shoulder?
Liver-related shoulder pain typically presents in the right shoulder, specifically the top of the shoulder or the shoulder blade area. This referred pain occurs because the liver sits beneath the right diaphragm, and the phrenic nerve, which innervates the diaphragm, shares nerve roots (C3-C5) with sensory nerves that supply the shoulder.
When a hepatic tumor or significant inflammation irritates the diaphragm, the brain misinterprets the signal as originating from the shoulder. Patients describe it as a dull ache or pressure that does not respond to movement or typical musculoskeletal treatments.
In cirrhosis communities, right shoulder pain queries spike among users who also discuss worsening fatigue and abdominal distension. This triad, shoulder pain plus fatigue plus ascites concern, appears in VIOLET's signal mapping as a high-specificity indicator of hepatic decompensation or new mass effect. No clinical guideline captures this behavioral pattern. No EHR codes for it.
Can you live 20 years with cirrhosis?
Yes, but the answer depends almost entirely on disease stage at detection and ongoing management. Patients with compensated cirrhosis (Child-Pugh A) have a median survival exceeding 12 years. Some patients diagnosed early with well-managed etiology, such as sustained virological response in hepatitis C or complete alcohol cessation, can live 20 years or longer.
Decompensated cirrhosis changes the picture dramatically. Median survival drops to 2 to 4 years without transplantation. The transition from compensated to decompensated disease is the critical inflection point, and it is precisely the window where behavioral signals are most abundant and most underutilized.
A 2018 MedPageToday review on telemedicine in chronic liver disease noted the potential for remote monitoring to optimize care delivery for this population. But remote monitoring captures what clinicians ask about. Behavioral intelligence captures what patients worry about, search for, and discuss with peers. These are fundamentally different data streams, and only one of them exists before the patient contacts the health system.
Personality changes and memory loss: the hepatic encephalopathy signal
Related searches for "personality changes liver cirrhosis" and "cirrhosis and memory loss" reveal a behavioral pattern that maps to subclinical hepatic encephalopathy (SHE). An estimated 20% to 80% of cirrhosis patients have some degree of SHE, depending on the testing method used. Most are never formally diagnosed.
In cirrhosis forums, caregiver posts about personality changes outnumber patient self-reports by roughly 4 to 1. Caregivers describe irritability, apathy, sleep-wake cycle reversal, and difficulty with routine tasks. These threads often appear 2 to 6 months before patients seek hepatology follow-up for worsening encephalopathy.
This behavioral signal has direct clinical relevance. Worsening hepatic encephalopathy in a cirrhosis patient can indicate declining hepatic reserve, portal hypertension progression, or new hepatic mass, all of which increase HCC risk. The caregiver behavioral signal is, in many cases, more timely than the patient's own clinical contact.
VIOLET captures these signals not as individual data points but as temporal patterns. A single search for "liver brain fog" means little. A cirrhosis community user who searches "liver brain fog," then "right side pain under ribs," then "cirrhosis cancer risk" over a 90-day window represents a fundamentally different risk profile.
Key statistics
Why current liver cancer surveillance misses these signals
The AASLD surveillance protocol relies on ultrasound and alpha-fetoprotein (AFP) testing every six months. This protocol has two structural weaknesses.
First, it requires patients to show up. The 24% adherence rate means three out of four cirrhosis patients are not screened on schedule. Behavioral data does not require a clinic visit. It generates continuously.
Second, ultrasound sensitivity for early HCC is only 47% as a standalone test (63% when combined with AFP). Even when patients are screened, small tumors are frequently missed. Behavioral signals do not replace imaging, but they can identify patients who need more aggressive surveillance before the next scheduled scan.
The gap between what surveillance recommends and what patients actually receive is a data problem. Not a data quantity problem. A data type problem. Clinical systems capture labs, imaging, and visit notes. They do not capture the search a cirrhosis patient runs at 2 a.m. about whether their shoulder pain means cancer.
Hepatocellular carcinoma behavioral signals as risk intelligence
The concept of hepatocellular carcinoma behavioral signals is distinct from clinical risk stratification. Clinical models like GALAD (Gender, Age, L3 fraction, AFP, DCP) and aMAP (age, Male sex, Albumin-Bilirubin, Platelets) predict HCC risk based on biomarkers and demographics. They are useful but static. They score patients at the time of a clinical encounter.
Behavioral signals are dynamic. They update in real time as patients interact with health information, peer communities, and search engines. When a cirrhosis patient who has been stable for two years suddenly begins searching for "liver cancer survival rates" and posting in HCC-specific forums, that behavioral shift carries prognostic weight that no biomarker panel captures in between visits.
This is what liver cancer risk intelligence means in practice. Not replacing clinical risk models, but supplementing them with a continuous signal that exists outside the four walls of a health system.
The data trust requirement for behavioral health intelligence
Behavioral data from cirrhosis communities is sensitive. Substance use disorder history, mental health symptoms, and liver disease stigma all intersect in this population. Using this data responsibly requires more than HIPAA compliance.
SuperTruth's Data Trust Index scores every record across eight dimensions: Provenance (25%), Consent (20%), Recency (15%), Quality (10%), Concordance (10%), Validation (10%), Breadth (5%), and Stability (5%). For behavioral signals derived from patient communities, Consent and Provenance carry the highest weight. Where did this signal originate? Was it generated in a context where the user had meaningful notice about data use? Can the chain of custody be verified?
These are not theoretical questions. The substance use disorder data challenge is directly relevant here, as many cirrhosis patients have alcohol use disorder histories that require 42 CFR Part 2 protections. Behavioral intelligence that touches this population needs consent architecture that accounts for the most restrictive applicable standard.
From behavioral signal to clinical action
The value of liver cancer data from cirrhosis communities is not in individual predictions. It is in population-level risk stratification that helps health systems allocate surveillance resources more effectively.
Consider a hepatology practice with 2,000 cirrhosis patients. AASLD guidelines say all 2,000 need biannual ultrasound. In practice, fewer than 500 are screened on time. Behavioral intelligence can identify the 200 patients showing escalating risk signals, the ones searching for new symptoms, posting about cognitive changes, and querying shoulder pain, and prioritize them for outreach.
This is not a replacement for universal screening. It is triage for a system that cannot deliver universal screening today. And it requires data that is scored, trusted, and governed. Not scraped.
VIOLET maps these signals across oncology populations. For liver cancer specifically, the platform tracks search behavior, community interaction patterns, and temporal signal escalation across the cirrhosis-to-HCC continuum. The result is not a diagnosis. It is intelligence that tells a health system where to look next.
VIOLET maps behavioral signals across 750+ oncology search terms before patients reach a clinic. If your team is working on cohort identification, trial recruitment, or oncology market intelligence for hepatocellular carcinoma, contact Louis Simeonidis at louis@supertruth.ai or (215) 918-4140.
Further reading:
Jason Alan Snyder
Co-founder of SuperTruth and Artists & Robots, and an inventor on the Data Trust Index patents. Twenty-plus years building technology inside Interpublic Group. He writes here nearly every day on data trust, provenance, and what AI should be allowed to act on, and publishes essays on his Substack.
About SuperTruth · LinkedIn · Substack · jasonalansnyder.com
See it in practice
750+ cancer search terms. Live in production.
VIOLET maps behavioral signals 12–18 months before clinical presentation.