Researchers created a computer program that can predict whether someone might develop coronary heart disease (a serious heart condition) by looking at simple health measurements like blood pressure and cholesterol levels. The program was tested on thousands of people’s health records and correctly identified who was at high risk about 83% of the time. The tool could help doctors catch heart disease early, when treatment is most effective. The researchers found that age, sex, high blood pressure, and high cholesterol are the strongest warning signs. A free online calculator based on this program is now available for anyone to use.

The Quick Take

  • What they studied: Can a computer program accurately predict who will develop heart disease using basic health information?
  • Who participated: The study used health records from hundreds of thousands of American adults collected through two major government health surveys (BRFSS and NHANES). The data included people of different ages, sexes, and health backgrounds.
  • Key finding: A machine learning program called LightGBM correctly predicted heart disease risk 83% of the time in initial testing and 85% of the time when tested on a completely different group of people. This was better than seven other computer programs tested.
  • What it means for you: If you’re concerned about heart disease risk, this tool could help you and your doctor identify if you need to make lifestyle changes or take preventive medications. However, this tool should complement, not replace, conversations with your doctor about your personal heart health.

The Research Details

Researchers took health information from a large government database of American adults and split it into two groups: a training group (70% of the data) and a testing group (30% of the data). They used the training group to teach eight different computer programs to recognize patterns that predict heart disease. Each program learned which health measurements were most important for making predictions.

After training, they tested all eight programs on the testing group to see which one worked best. The winning program (LightGBM) was then tested again on a completely separate group of people from a different health survey to make sure it worked in the real world, not just on the training data.

Finally, the researchers used a special analysis technique called SHAP to figure out which health factors the computer program thought were most important for predicting heart disease.

This approach is important because it uses real-world health data from thousands of people rather than just laboratory studies. Testing on two different datasets (internal and external validation) helps ensure the tool actually works for different groups of people, not just the ones it was trained on. This makes doctors more confident that the tool will be helpful in real medical practice.

The study’s strengths include using large, representative datasets from government health surveys and testing the program twice on different groups of people. The researchers also compared eight different computer programs to find the best one. However, the paper doesn’t specify exactly how many people were in the study, and the results may not apply equally to all ethnic groups or people with very different health profiles than those in the surveys.

What the Results Show

The LightGBM program was the most accurate of the eight computer programs tested. When tested on new data it had never seen before, it correctly identified 80% of people who would develop heart disease (called sensitivity) and correctly identified 70% of people who would not develop it (called specificity). The program’s overall accuracy score was 0.825 on the first test and 0.851 on the second test with a different group—both very good scores.

To put this in perspective, a score of 0.5 would mean the program is just guessing like a coin flip, and a score of 1.0 would mean perfect accuracy. Scores above 0.8 are considered excellent in medical prediction tools.

The analysis showed that the four most important factors the program used were: your age, your sex, whether you have high blood pressure, and whether you have high cholesterol. These four factors alone explained most of the program’s predictions, which matches what doctors already know about heart disease risk.

The other seven computer programs tested also performed reasonably well, but none were as accurate as LightGBM. The study also confirmed that the program worked similarly well when tested on the second, independent group of people, which suggests it should work reliably in real-world medical settings. The SHAP analysis revealed that the program’s predictions were based on logical, understandable factors—not mysterious patterns that doctors couldn’t explain.

This research builds on decades of medical knowledge showing that age, sex, high blood pressure, and high cholesterol are major heart disease risk factors. The new contribution is using modern computer programs to combine these factors in a more sophisticated way than traditional risk calculators. The accuracy rates (0.825-0.851) are comparable to or better than existing heart disease risk prediction tools that doctors currently use.

The study has several important limitations. First, the researchers didn’t clearly report the total number of people included, making it hard to assess how large the study truly was. Second, the data came from American health surveys, so the results may not apply equally to people in other countries with different genetics or healthcare systems. Third, the study only looked at people’s health information at one point in time, rather than following them over years to see who actually developed heart disease. Finally, the program’s accuracy might be different for certain ethnic groups or age ranges, but this wasn’t thoroughly examined.

The Bottom Line

If you have risk factors for heart disease (high blood pressure, high cholesterol, family history, or are over 40), consider using this tool with your doctor to assess your risk level. The tool appears to be reliable for identifying who needs closer monitoring or preventive treatment. However, this should be one part of a complete heart health evaluation, not a replacement for medical advice. Confidence level: Moderate to High for general risk assessment.

This tool is most useful for adults concerned about heart disease risk, people with family histories of heart disease, and healthcare providers looking for better screening methods. It may be less useful for very young people or those already diagnosed with heart disease who need different management strategies. People should not use this tool to diagnose themselves or avoid seeing a doctor.

If you make lifestyle changes based on this tool’s recommendations (like exercising more, eating healthier, or taking medications), you might see improvements in blood pressure and cholesterol within 3-6 months. However, actual prevention of heart disease takes years of consistent healthy habits. This tool is meant for early identification, not immediate results.

Want to Apply This Research?

  • Track your blood pressure, cholesterol levels, and weight monthly. Input these measurements into the app alongside your age and sex to get updated risk assessments every 3 months. This shows whether your risk is improving with lifestyle changes.
  • Use the app’s risk score as motivation to set specific goals: if the tool shows you’re at moderate-to-high risk, commit to 150 minutes of exercise weekly, reduce salt intake, and schedule a doctor’s visit to discuss your results and medication options if needed.
  • Set quarterly reminders to re-enter your health measurements and check your updated risk score. Share results with your doctor to track progress over time. If your score improves, it reinforces that your lifestyle changes are working. If it worsens, it’s a signal to intensify efforts or adjust medications.

This research describes a computer tool for assessing heart disease risk based on health data. This tool is not a substitute for professional medical diagnosis or treatment. Always consult with a qualified healthcare provider before making decisions about your health, medications, or lifestyle changes. The tool’s predictions are estimates based on population data and may not apply equally to all individuals. If you experience chest pain, shortness of breath, or other heart attack symptoms, seek emergency medical care immediately rather than relying on this tool.