According to research reviewed by Gram, a Random Forest computer model can estimate a person’s risk of losing all their natural teeth using age, education, smoking, income, and diet. In a 2026 study of 9,230 US adults, about 7.86% had lost all their teeth, and low fiber and magnesium intake were linked to higher risk.
A new study published in Medicine, reviewed by Gram Research analysis, looked at why some adults lose all of their natural teeth. Researchers used national health survey data from 9,230 American adults aged 20 and older and built computer models to predict who was most at risk. About 7.86% of participants had no natural teeth left. Age, years of education, smoking, and household income level were the strongest predictors. Diet also appeared to matter. People who ate less fiber and magnesium, and more sugar, tended to have higher risk. The best-performing model, called Random Forest, was good at sorting people by risk and its predicted probabilities appeared reliable. Because the study looked at a snapshot in time, it cannot prove that any food causes tooth loss. Still, tools like this may help dentists and health teams spot high-risk adults early and guide nutrition advice.
Key Statistics
A 2026 cross-sectional study of 9,230 US adults aged 20 and older, published in Medicine, found that about 7.86% of participants had lost all of their natural teeth.
In the same 2026 study, the gradient boosting models XGBoost and LightGBM achieved an AUC score of 0.855 for predicting complete tooth loss, indicating good ability to sort people by risk.
The 2026 study reported that higher intakes of dietary fiber and magnesium were associated with lower risk of complete tooth loss among US adults, while higher total sugar intake was associated with higher risk.
The Quick Take
- What they studied: Researchers asked whether a computer model could predict which adults are likely to lose all of their natural teeth, and which factors, including diet, were most closely linked to that risk.
- Who participated: The study used survey data from 9,230 US adults aged 20 and older. About 7.86% of them had lost all of their natural teeth. The data came from a large national health and nutrition survey run by the US government.
- Key finding: Age, years of education, smoking, and household income were the strongest predictors of complete tooth loss. Lower intake of fiber and magnesium was linked to higher risk, while higher sugar intake was linked to higher risk too. The computer models sorted people by risk with good accuracy, though not perfectly.
- What it means for you: Eating more fiber-rich foods, such as beans, vegetables, whole grains, and fruit, and including magnesium-rich foods like nuts, seeds, and leafy greens, may be a sensible habit for your overall health. Keeping up with regular dental checkups and quitting smoking are also strongly supported steps. This study alone cannot prove that diet directly prevents tooth loss.
The Research Details
This was a cross-sectional study, which means researchers took a snapshot of many people at one point in time instead of following them over several years. They used health survey data that included background details like age, income, and education, along with information about chronic illnesses, smoking, and what people ate. The team started with 44 possible factors and used a statistical method called LASSO to narrow the list down to 11 of the most useful ones.
The researchers then trained five different computer models to predict tooth loss. They split the data so that 70% was used to teach the models and 30% was saved to test how well they worked on people they had not seen before. This approach helps show whether a model can handle new information rather than just memorizing the data it learned from.
To understand the results, the team used a tool called SHAP, which shows how much each factor pushed a prediction up or down. They also used statistical curves to check whether the link between diet and tooth loss was a straight line or a more complicated pattern. This helped them see, for example, that fiber and magnesium appeared protective across a range of intake levels.
A snapshot study is a good first step for finding patterns across large groups of people. Because the sample was large and came from a national survey, the findings reflect a wide range of US adults. Testing the models on a separate group of people also gives a more honest measure of how well a tool might work in real life.
However, a snapshot study cannot show what caused what. It can show that two things tend to occur together, but it cannot prove that eating more fiber prevents tooth loss. Confirming a cause would require long-term studies or controlled trials.
The study used a large sample, a standard national survey, and a clear comparison between training and testing data, all of which are strengths. The researchers checked several measures of model performance, including how accurate the predicted probabilities were, which is a good sign of reliability.
The paper also has some issues readers should know about. The title refers to survey years from 2009 to 2014, but the text describes data from 2005 to 2018, so the exact time period is unclear. Dietary information in surveys is often self-reported, which can be inaccurate. The study has not yet been repeated in other populations, so its predictions should be treated as early evidence.
What the Results Show
The study found that about 7.86% of US adults aged 20 and older had lost all of their natural teeth. Among the 11 factors the models relied on most, age, years of education, smoking status, and household income-to-poverty ratio stood out as the strongest predictors. Older adults, people with less education, smokers, and those with lower incomes were more likely to have lost all their teeth.
Diet showed a clear pattern as well. Higher intakes of dietary fiber and magnesium were associated with lower risk of tooth loss, and the relationship appeared to level off at higher intake levels. Total sugar intake showed a link to higher risk, and this relationship also followed a curved pattern rather than a simple straight line.
In terms of prediction, two gradient boosting models, XGBoost and LightGBM, achieved the highest accuracy at sorting people by risk, with an AUC score of 0.855 (a score of 1.0 would be perfect and 0.5 would be no better than guessing). The researchers ultimately chose Random Forest because it balanced accuracy, reliable predicted probabilities, and practical usefulness in decision-making.
The models also showed that predicted risk levels matched real outcomes well, with a very low calibration error of 0.0097. This matters because a tool that gives a 30% risk estimate should be right about roughly 30% of similar people. The analysis also suggested that adults with diabetes or heart disease had higher risk, which fits with the broader idea that overall health and tooth health are connected.
The researchers described a pattern of older age, lower income, lower antioxidant intake, and more chronic conditions in people who had lost their teeth. This suggests tooth loss often reflects a combination of life circumstances rather than one single cause.
Previous research has consistently linked tooth loss with older age, smoking, lower income, and diabetes, so this study’s main risk factors are not surprising. Earlier studies have also connected diets rich in fiber and minerals with better oral and overall health. This study adds value by combining many of these factors into a single prediction model, which is a newer approach for this topic.
Because this was a snapshot study, it can show associations but cannot prove that diet, smoking, or income directly cause tooth loss. Diet information was likely self-reported, which can be less accurate than measured data. The paper also describes survey years inconsistently, with the title citing 2009 to 2014 and the text citing 2005 to 2018, which makes the exact time period unclear. The models were tested within one national dataset and have not yet been checked in other countries, clinics, or groups of people. Finally, the study did not account for every possible factor, such as detailed brushing habits or dental care access, which may also play a role.
The Bottom Line
Based on this study, it is reasonable to aim for a diet with plenty of fiber and magnesium, while limiting added sugar, as part of general healthy eating. Confidence in these diet suggestions is moderate, since the link is consistent with other research but this study cannot prove cause and effect. Confidence is higher for the advice to avoid smoking and to keep up with regular dental visits, since these are well-established steps for protecting teeth.
The computer model itself is best viewed as an early screening idea rather than a finished clinical tool. It would need further testing before being used in routine care.
Older adults, people who smoke, and those with lower incomes or limited formal education may find this research especially relevant, since these groups were at higher predicted risk. Dentists, dental hygienists, and public health planners may be interested in the screening approach. Adults who already have healthy teeth and a balanced diet may not need to change much based on this study alone. People with specific medical conditions, such as kidney disease or those who take medications that affect minerals, should talk with a doctor before making major diet changes.
Diet changes usually take weeks to months to show effects on gum health and overall oral health, and tooth loss itself develops over many years. Stopping smoking may begin to improve gum health within a few months, while long-term benefits build over years. Realistic expectations are that better eating habits support oral health gradually, rather than producing quick or dramatic results.
Frequently Asked Questions
What factors are most linked to losing all your natural teeth?
A 2026 study of 9,230 US adults found age, years of education, smoking status, and household income were the strongest predictors of complete tooth loss. Lower fiber and magnesium intake and higher sugar intake were also linked to higher risk.
Can a computer model predict who will lose their teeth?
Researchers built several computer models, and Random Forest was selected as the most practical. The models sorted people by risk with good accuracy in testing, but they are early tools and still need validation in other populations before routine clinical use.
Does eating more fiber help prevent tooth loss?
Fiber intake was associated with lower risk of complete tooth loss in a large US survey analysis. However, this snapshot study cannot prove cause and effect, so fiber should be viewed as part of a healthy diet rather than a guaranteed way to keep teeth.
How common is complete tooth loss among American adults?
In a 2026 analysis of 9,230 US adults aged 20 and older, about 7.86% had lost all of their natural teeth. The risk was higher among older adults, smokers, and people with lower income or less formal education.
Want to Apply This Research?
- Track daily grams of dietary fiber and estimated magnesium intake, along with added sugar grams, and aim for steady progress toward a fiber goal such as 25 to 30 grams per day. Log whether you brushed twice a day and whether you had a dental checkup in the past 12 months.
- Add one fiber-rich food to each meal, such as beans, lentils, berries, oats, or vegetables. Swap one sugary snack per day for a handful of nuts or seeds, which provide magnesium. Use the app’s reminder feature to prompt a dental checkup every six months.
- Review fiber, magnesium, and sugar trends weekly and look for patterns over several months rather than single days. Record dental visits and any changes in gum comfort or sensitivity. Share summary reports with a dentist or doctor at routine appointments so they can factor diet and habits into their advice.
This article summarizes a single cross-sectional study that shows associations, not proof of cause and effect. It is for general education only and does not replace advice from a dentist, doctor, or registered dietitian. Anyone with concerns about tooth loss, gum health, or diet should consult a qualified health professional before making changes.
This research translation is published by Gram Research, the science division of Gram, an AI-powered nutrition tracking app.