A simple blood test called the glucose-ketone index (GKI), measured on the first day of ketogenic diet therapy, can predict whether the diet will reduce seizures by at least 50% in children with refractory epilepsy. According to Gram Research analysis of 189 children, a GKI above 9.63 on day one indicated a higher likelihood of poor response to treatment, with 63% accuracy. This early predictor could help doctors decide within days whether to continue the diet or try alternative treatments.
Researchers studied 189 children with hard-to-treat epilepsy who tried a special ketogenic diet. They discovered that a simple blood test measuring the ratio of glucose to ketones—called the GKI—taken on the first day of treatment could predict whether the diet would reduce seizures by at least half. A GKI score above 9.63 suggested the diet might not work well for that child. This finding could help doctors decide quickly whether to continue the diet or try a different treatment, potentially saving time and helping children get better seizure control faster.
Key Statistics
A 2026 study of 189 children with refractory epilepsy found that the glucose-ketone index (GKI) measured on the first day of ketogenic diet therapy predicted treatment response with 63% accuracy, with a cutoff value of 9.63.
Research reviewed by Gram on 189 children showed that combining the glucose-ketone index with blood glucose levels (0.75 mmol/L) improved the predictive accuracy for ketogenic diet efficacy from 0.638 to 0.661.
A 2026 analysis of 189 children found that the early metabolic transition period—from two days before starting the ketogenic diet through the first day of treatment—represents a critical window for predicting whether the diet will reduce seizures by at least 50%.
The Quick Take
- What they studied: Can a simple blood test predict whether a special high-fat, low-carb diet will help children with epilepsy that doesn’t respond to regular medicines?
- Who participated: 189 children with refractory epilepsy (seizures that don’t stop with medication) who started a classic ketogenic diet between 2018 and 2024
- Key finding: According to Gram Research analysis, a blood measurement called the glucose-ketone index (GKI) taken on day one of the diet could predict treatment success. A GKI above 9.63 suggested the diet would likely not reduce seizures by at least 50%, with 63% accuracy.
- What it means for you: If your child has epilepsy that doesn’t respond to medicines, doctors could use this blood test early on to determine if the ketogenic diet is worth continuing or if another treatment should be tried instead. This could save time and help find the right treatment faster.
The Research Details
Researchers looked back at medical records from 189 children with refractory epilepsy who received a classic ketogenic diet (a diet very high in fat and very low in carbohydrates) between 2018 and 2024. They measured three things in the children’s blood at different time points: glucose (blood sugar), ketones (a fuel the body makes when eating very few carbs), and the GKI (the ratio of glucose to ketones). They tracked these measurements starting two days before the diet began, through the first day of the diet, and continuing for 90 days.
After three months of treatment, the researchers divided the children into two groups: those whose seizures decreased by at least 50% (responders) and those whose seizures didn’t decrease that much (non-responders). They then compared the blood measurements between these two groups to see which measurements best predicted who would respond well to the diet.
The researchers used a statistical tool called ROC analysis to find the best cutoff values—the specific numbers that would best separate children who would benefit from the diet from those who wouldn’t.
This research approach is important because it looks at real-world data from actual patients rather than just testing the diet in a controlled lab setting. By measuring blood markers at multiple time points, the researchers could identify exactly when and which measurements were most predictive. Finding an early predictor is valuable because it means doctors don’t have to wait three months to know if the diet is working—they could know within the first day.
This study has some strengths: it included a reasonably large sample size (189 children), tracked patients over a long time period (2018-2024), and measured multiple related markers. However, it’s a retrospective study, meaning researchers looked back at existing medical records rather than following children forward in a planned way. This type of study can’t prove cause-and-effect as strongly as a randomized controlled trial. The study was conducted at what appears to be a single center, so results may not apply equally to all populations. The predictive accuracy (63%) is moderate, not extremely high, meaning the test won’t perfectly predict outcomes for every child.
What the Results Show
On the very first day of ketogenic diet therapy, the glucose-ketone index (GKI) was the single best predictor of whether the diet would work, with an accuracy measure (AUC) of 0.638. The optimal cutoff value was 9.63—children with a GKI above this number were more likely to not respond well to the diet. When researchers combined the GKI with a specific blood glucose level (0.75 mmol/L), the prediction accuracy improved slightly to 0.661.
Interestingly, the early transition period—from two days before starting the diet through the first day of actual ketogenic diet therapy—appeared to be a critical window. During this time, the body is shifting from using glucose as fuel to using ketones. The researchers found that monitoring how quickly and dramatically the GKI changed during this period could help doctors understand whether a child’s body was responding appropriately to the diet.
Children who showed a GKI greater than 9.63 on day one had a higher likelihood of being non-responders, meaning their seizures would not decrease by at least 50%. This suggests that some children’s bodies may not be metabolically suited to benefit from the ketogenic diet, or that they may need different dosing or timing adjustments.
The study found that blood ketone levels and blood glucose levels individually were less predictive than the GKI ratio. This suggests that it’s not just about how high the ketones go or how low the glucose drops—it’s the balance between them that matters. The timing of measurements also mattered significantly; measurements taken later in the treatment course (after day one) were less predictive of final outcomes. This reinforces that the first day is a critical window for assessment.
Previous research has shown that the ketogenic diet can help children with refractory epilepsy, but doctors have struggled to predict in advance which children will benefit. This study adds to that knowledge by identifying a specific, measurable marker that can be checked early. Other studies have looked at ketone levels or glucose levels separately, but this research emphasizes the importance of their ratio (the GKI) rather than either measure alone. The finding that the early metabolic transition period is critical aligns with what we know about how the ketogenic diet works—it requires the body to shift its fuel source, and how well and quickly that happens may determine effectiveness.
The study has several important limitations. First, it’s retrospective, meaning researchers looked backward at existing records rather than prospectively following children forward with a planned protocol. This can introduce bias based on which records were kept and how complete they were. Second, the predictive accuracy of 63% is moderate—it’s better than flipping a coin, but it won’t correctly predict outcomes for all children. Third, the study appears to have been conducted at a single medical center, so the results may not apply equally to children in different geographic regions or healthcare settings with different treatment protocols. Fourth, the study doesn’t explain why some children have a high GKI on day one or what factors influence this. Finally, the study doesn’t tell us whether adjusting the diet based on GKI measurements actually improves outcomes—it only shows that GKI predicts outcomes.
The Bottom Line
For children with refractory epilepsy starting a ketogenic diet, doctors should consider measuring the glucose-ketone index (GKI) on the first day of treatment. A GKI above 9.63 suggests the child may not respond well to the diet, which could prompt doctors to discuss alternative treatments or dietary adjustments earlier rather than waiting three months. However, this test should not be the only factor in deciding whether to continue the diet—individual response varies, and some children with higher GKI values may still benefit. Confidence in this recommendation is moderate, as the test’s accuracy is around 63%.
This research is most relevant for: (1) children with refractory epilepsy and their families considering the ketogenic diet, (2) neurologists and pediatricians treating children with hard-to-control seizures, and (3) dietitians specializing in medical nutrition therapy for epilepsy. This research is less relevant for children whose seizures are well-controlled with medication or for adults with epilepsy, as the study only included children.
The GKI measurement can be done on day one of treatment, providing early information within days rather than weeks. However, the full three-month trial period is still recommended to assess actual seizure reduction, as the GKI is a predictor, not a definitive outcome measure. Some children may show delayed responses to the diet, so a full assessment period is important before making final decisions.
Frequently Asked Questions
Can a blood test predict if the ketogenic diet will stop my child’s seizures?
A blood test measuring the glucose-ketone index (GKI) on day one of ketogenic diet therapy can help predict response with about 63% accuracy. A GKI above 9.63 suggests lower likelihood of at least 50% seizure reduction, but individual responses vary and should be confirmed over three months.
What is the glucose-ketone index and why does it matter for epilepsy treatment?
The glucose-ketone index (GKI) is the ratio of blood glucose to blood ketones. It matters because it reflects how well your body is transitioning from using sugar to using ketones for fuel—a key mechanism of the ketogenic diet. A balanced GKI may indicate better metabolic adaptation to the diet.
How soon can doctors tell if the ketogenic diet will work for my child’s seizures?
Doctors can get an early indication within the first day using the GKI blood test, but full assessment requires three months of treatment to measure actual seizure reduction. The early test helps guide whether to continue, adjust, or consider alternative treatments sooner.
What should I do if my child’s GKI is high on the first day of the diet?
A high GKI (above 9.63) suggests the diet may not work optimally, but doesn’t mean it won’t help. Discuss with your doctor about potential adjustments to diet composition, stricter carbohydrate reduction, or monitoring for delayed response before deciding to stop.
Is the glucose-ketone index test available at all hospitals?
The GKI requires measuring both blood glucose and blood ketones, which are standard tests available at most hospitals and labs. However, not all doctors may be familiar with using GKI specifically for predicting ketogenic diet response, so discuss this with your child’s neurologist.
Want to Apply This Research?
- Track the glucose-ketone index (GKI) daily for the first 90 days of ketogenic diet therapy. Record blood glucose and blood ketone measurements at the same time each day, calculate the GKI (glucose ÷ ketones), and log the result. Also track seizure frequency and severity daily to correlate metabolic changes with seizure control.
- Work with your healthcare team to establish a regular blood testing schedule during the first week of the diet. Use the app to set reminders for testing times, record results immediately, and share trends with your doctor. If your GKI is above 9.63 on day one, discuss with your doctor whether diet adjustments (like stricter carb reduction or different fat ratios) might help optimize your metabolic response.
- Create a dashboard showing GKI trends over the first 90 days alongside seizure frequency. Set weekly check-in reminders to review whether GKI is trending downward (which may correlate with better seizure control). If GKI remains high after one week despite dietary adherence, flag this for discussion with your healthcare team about potential adjustments or alternative treatments.
This research provides information about predicting ketogenic diet response in children with refractory epilepsy but should not replace professional medical advice. The glucose-ketone index is a predictive tool with moderate accuracy (63%), not a definitive diagnostic test. Any decisions about starting, continuing, or stopping the ketogenic diet should be made in consultation with your child’s neurologist or epilepsy specialist. Blood glucose and ketone monitoring should only be performed under medical supervision. Individual responses to the ketogenic diet vary significantly, and some children with higher GKI values may still benefit from the diet. This study was conducted in a specific population and may not apply equally to all children or healthcare settings.
This research translation is published by Gram Research, the science division of Gram, an AI-powered nutrition tracking app.