According to Gram Research analysis, an artificial intelligence system can predict gallstone disease with about 80% accuracy using blood tests and patient health data. The study, published in PLOS One in 2026, found that C-Reactive Protein (CRP) and Vitamin D levels were the strongest predictors of gallstone risk. While this technology is still experimental, it suggests blood tests could eventually help doctors identify gallstones before symptoms appear, which is important since 80% of people with gallstones don’t realize they have them.
Researchers have developed an artificial intelligence system that can predict whether someone has gallstones by analyzing blood test results and other patient data. The study, published in PLOS One, tested a machine learning model called CatBoost that achieved about 80% accuracy in identifying gallstone disease. Interestingly, two specific markers—C-Reactive Protein (CRP) and Vitamin D levels—were the strongest predictors of gallstone risk. This breakthrough could help doctors identify gallstone disease earlier, especially since most people with gallstones don’t realize they have them. The AI system was designed to be transparent, showing doctors exactly which factors influenced each prediction.
Key Statistics
A 2026 research article published in PLOS One found that a machine learning system achieved 80.42% accuracy in predicting gallstone disease using 19 key health measurements selected from patient blood tests and medical data.
According to the 2026 PLOS One study, C-Reactive Protein (CRP) and Vitamin D were identified as the most influential predictors of gallstone disease across multiple artificial intelligence explanation methods.
In the best-performing test of the AI system, researchers achieved 87.50% accuracy in identifying gallstone disease, demonstrating that the technology can perform even better under optimal conditions.
The research showed that the optimized AI system maintained balanced performance across different accuracy metrics, with precision of 81.97% and recall of 77.76%, indicating reliable detection of both true gallstone cases and non-cases.
The Quick Take
- What they studied: Can an artificial intelligence system predict gallstone disease by analyzing patient blood tests and medical information?
- Who participated: The study used publicly available patient data with 38 different health measurements. The exact number of patients wasn’t specified in the research, but the data included various blood markers and health indicators.
- Key finding: The AI system correctly identified gallstone disease about 80% of the time, with some test runs reaching 87% accuracy. The most important predictors were inflammation markers (CRP) and Vitamin D levels.
- What it means for you: This research suggests doctors might soon use simple blood tests to screen for gallstones before symptoms appear. Since 80% of people with gallstones have no symptoms, this could catch the disease earlier. However, this is still experimental technology and shouldn’t replace current diagnostic methods yet.
The Research Details
Researchers created a computer program trained to recognize patterns in patient health data that indicate gallstone disease. They used a machine learning technique called CatBoost, which is good at finding patterns in medical information. To make sure their results were reliable, they tested the system multiple times using a method called 5-fold cross-validation—imagine splitting your data into 5 groups, training on 4 groups, and testing on the 5th group, then repeating this process 5 different ways.
They also used something called the Sea Lion Optimization Algorithm (SLOA) to improve their AI system by selecting the most important health markers. This reduced the number of measurements needed from 38 down to 19, making the system simpler and faster while keeping it accurate.
Finally, they applied special techniques called SHAP, LIME, and DiCE to make the AI’s decisions transparent. These tools show doctors exactly which patient factors influenced each prediction, rather than treating the AI like a black box.
Most gallstone research focuses on using imaging like ultrasounds, but this study looked at whether simple blood tests could work instead. Blood tests are cheaper, faster, and more accessible than imaging. Making the AI transparent is also crucial—doctors need to understand why the system makes predictions before they can trust it in real medical practice.
The study was published in PLOS One, a reputable peer-reviewed journal. The researchers used proper statistical methods and tested their system multiple times to ensure reliability. However, the exact sample size wasn’t clearly reported, which makes it harder to judge how well these results might apply to different populations. The study used publicly available data, which is good for reproducibility but may not represent all patient types equally.
What the Results Show
The CatBoost AI system achieved an average accuracy of 79.58% when identifying gallstone disease using 38 different patient measurements. This means that roughly 8 out of every 10 predictions were correct. In the best-performing test (fold-1), the system reached 86.46% accuracy, showing it can perform even better under certain conditions.
When researchers simplified the system to use only 19 key measurements selected by the optimization algorithm, accuracy improved slightly to 80.42%. This is important because it shows you don’t need all 38 measurements—just the most important ones work nearly as well.
The system’s performance was balanced across different metrics. The F1-score (which measures overall accuracy) was around 79-80%, precision (how often it correctly identifies gallstones when it says they’re present) was about 81%, and recall (how many actual gallstone cases it catches) was about 77%. This balance means the system is reliable for both identifying disease and ruling it out.
The most important discovery was identifying which health markers matter most. C-Reactive Protein (CRP)—a blood test that measures inflammation—and Vitamin D levels were consistently the strongest predictors of gallstone disease across all the AI explanation methods tested. This finding aligns with previous medical research suggesting inflammation and vitamin D deficiency may play roles in gallstone formation. The fact that multiple explanation techniques identified the same key factors adds confidence to this finding.
Previous research on gallstone detection focused mainly on imaging-based machine learning (ultrasounds and CT scans), which achieved high accuracy but requires expensive equipment and trained technicians. This study is novel because it shows that tabular data (simple blood tests and patient information) can also predict gallstones effectively. The 80% accuracy is comparable to many imaging-based systems while being much more practical for screening. The emphasis on explainable AI also represents a shift in medical research toward transparency, addressing a major criticism of earlier ‘black box’ AI systems.
The study didn’t clearly report the total number of patients included, making it difficult to assess whether results would apply to larger or more diverse populations. The data came from a single public dataset, which may not represent all types of patients equally—for example, it might not include enough data from different age groups or ethnic backgrounds. The study was purely computational and didn’t test the system in real clinical settings with actual doctors and patients. Additionally, the research doesn’t explain why CRP and Vitamin D are important predictors, only that they are. Finally, achieving 80% accuracy means 20% of predictions are wrong, which is too high for the system to be used alone without doctor confirmation.
The Bottom Line
This research is promising but still experimental. It suggests that blood tests measuring CRP and Vitamin D, combined with AI analysis, might help identify gallstone risk. However, current medical practice should not change based on this single study. Doctors should continue using proven diagnostic methods like ultrasound. This technology might eventually become useful for screening people at risk, but more testing in real clinical settings is needed first. If you have gallstone symptoms or risk factors, consult your doctor about appropriate screening methods.
This research is most relevant to medical researchers, AI developers, and healthcare technology companies working on diagnostic tools. Doctors might eventually use this technology for screening patients at risk of gallstones, particularly those with elevated inflammation or low vitamin D. People with family histories of gallstones or those with risk factors (obesity, certain diets) might benefit from future screening tools based on this research. However, people without symptoms or risk factors don’t need to change their behavior based on this study.
This is early-stage research. It typically takes 5-10 years for a new diagnostic technology to move from research to clinical use. Before that happens, the system needs testing on larger patient populations, validation in real hospitals, and regulatory approval. Don’t expect this AI tool to be available in your doctor’s office immediately, but it represents progress toward better gallstone screening.
Frequently Asked Questions
Can blood tests alone diagnose gallstones?
Not yet. This research shows that AI analyzing blood tests can predict gallstone risk with about 80% accuracy, but this is still experimental. Doctors currently use ultrasound imaging for definitive diagnosis. Blood tests measuring CRP and Vitamin D may eventually help screen for risk, but they cannot replace imaging confirmation.
What are the most important blood markers for gallstone risk?
According to the 2026 PLOS One study, C-Reactive Protein (CRP), which measures inflammation, and Vitamin D levels were the strongest predictors of gallstone disease. These two markers consistently ranked as most important across all analysis methods tested.
How accurate is this AI system for detecting gallstones?
The AI system achieved an average accuracy of 80.42% in identifying gallstone disease, with some tests reaching 87.50% accuracy. However, this means about 20% of predictions are incorrect, so it cannot be used alone without doctor confirmation and imaging tests.
When will this AI gallstone detection tool be available to patients?
This is still early-stage research. Typically, new diagnostic technologies take 5-10 years to move from research to clinical use. The system needs testing on larger patient populations and validation in real hospitals before doctors can use it in practice.
Should I get tested for gallstones if I have low vitamin D?
Low Vitamin D may be associated with gallstone risk according to this research, but it’s not a definitive indicator. If you have symptoms (upper right abdominal pain, nausea after fatty meals) or risk factors (obesity, family history), discuss screening with your doctor. Otherwise, maintain healthy Vitamin D levels for overall health.
Want to Apply This Research?
- Track your inflammation markers (CRP levels) and Vitamin D levels through regular blood tests. Record these values quarterly along with any gallstone symptoms (upper right abdominal pain, nausea after fatty meals) to identify patterns.
- If your CRP or Vitamin D levels are abnormal, work with your doctor on targeted improvements: increase vitamin D through sunlight exposure or supplements, and reduce inflammation through anti-inflammatory foods (fatty fish, leafy greens, berries). Log dietary changes and retest after 8-12 weeks.
- Set quarterly reminders for blood work to monitor CRP and Vitamin D trends. Create a health dashboard showing these values over time alongside any symptoms. Share this data with your doctor to assess gallstone risk and discuss preventive strategies.
This research describes an experimental artificial intelligence system and has not been validated for clinical use. It should not be used to diagnose, treat, or prevent gallstone disease. If you experience symptoms of gallstones (severe pain in the upper right abdomen, back pain between shoulder blades, or nausea), consult a healthcare provider immediately. Current diagnosis of gallstone disease requires imaging studies such as ultrasound, performed by qualified medical professionals. Always discuss any health concerns with your doctor before making medical decisions. This article is for educational purposes only and does not constitute medical advice.
This research translation is published by Gram Research, the science division of Gram, an AI-powered nutrition tracking app.